Upload fine-tuned Qwen3-VL-2B affordance model with dataset, checkpoints, and results

.gitattributes

@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+checkpoints/checkpoint-2000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoints/checkpoint-2145/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+results/inference_comparison.png filter=lfs diff=lfs merge=lfs -text
+results/training_curves.png filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,297 @@
+---
+library_name: peft
+base_model: unsloth/qwen3-vl-2b-instruct-unsloth-bnb-4bit
+license: apache-2.0
+tags:
+  - qwen3-vl
+  - affordance
+  - visual-affordance
+  - egocentric
+  - lora
+  - qlora
+  - fine-tuned
+  - vision-language
+  - unsloth
+language:
+  - en
+pipeline_tag: image-text-to-text
+datasets:
+  - sanskxr02/fva_affordance_v1
+---
+
+# Qwen3-VL-2B Affordance Fine-Tuned (QLoRA)
+
+A fine-tuned **Qwen3-VL-2B-Instruct** model for **visual affordance detection** from egocentric (first-person view) images. The model identifies what actions can be performed on visible objects in a scene.
+
+## Model Details
+
+| Property | Value |
+|---|---|
+| **Base Model** | [Qwen/Qwen3-VL-2B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct) |
+| **Method** | QLoRA 4-bit (NF4 quantization) |
+| **LoRA Rank** | 16 |
+| **LoRA Alpha** | 32 |
+| **LoRA Target Modules** | q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj |
+| **Trainable Parameters** | 17.4M / 1.52B (1.15%) |
+| **Training Framework** | [Unsloth](https://github.com/unslothai/unsloth) + TRL SFTTrainer |
+| **Training Hardware** | NVIDIA RTX 3060 Ti (8GB VRAM) |
+| **Training Time** | ~64 minutes |
+| **PEFT Version** | 0.18.1 |
+
+## Training Data
+
+Fine-tuned on [sanskxr02/fva_affordance_v1](https://huggingface.co/datasets/sanskxr02/fva_affordance_v1) - an egocentric visual affordance dataset containing:
+
+| Property | Value |
+|---|---|
+| **Total Images** | 3,177 frames |
+| **Train Split** | 2,859 (90%) |
+| **Validation Split** | 318 (10%) |
+| **Image Size** | 512 x 288 (16:9, resized to 384x216 during training) |
+| **Unique Affordances** | 377 |
+| **Task Categories** | food_preparation (coffee making), tool_use (wine bottling) |
+| **View** | Egocentric / First-person |
+
+### Affordance Examples
+`pour`, `pull bottle`, `pick up bottle`, `press button`, `pour coffee into cup`, `grasp coffee grinder`, `stir`, `hold bottle`, `pack bottles`, `put cup on counter`, etc.
+
+## Training Results
+
+| Metric | Value |
+|---|---|
+| **Initial Loss** | 3.4227 |
+| **Final Train Loss** | 0.0747 |
+| **Best Eval Loss** | 0.0976 |
+| **Loss Reduction** | 97.8% |
+| **Epochs** | 3 |
+| **Total Steps** | 2,145 |
+
+### Training Curves
+
+![Training Curves](results/training_curves.png)
+
+![Loss by Epoch](results/loss_by_epoch.png)
+
+### Inference Results
+
+![Inference Comparison](results/inference_comparison.png)
+
+## How to Use
+
+### Requirements
+
+```bash
+pip install torch==2.8.0 torchvision==0.23.0 --index-url https://download.pytorch.org/whl/cu126
+pip install unsloth transformers>=4.57.1 peft>=0.14.0 bitsandbytes>=0.49.1
+pip install accelerate qwen-vl-utils Pillow
+```
+
+### Quick Inference
+
+```python
+import os
+os.environ["TORCHDYNAMO_DISABLE"] = "1"  # Required on Windows
+
+from unsloth import FastVisionModel
+from qwen_vl_utils import process_vision_info
+
+# Load model with LoRA adapter
+model, tokenizer = FastVisionModel.from_pretrained(
+    "Kavin60606/qwen3-vl-2b-affordance-finetuned",
+    load_in_4bit=True,
+    max_seq_length=2048,
+)
+FastVisionModel.for_inference(model)
+
+# Prepare input
+messages = [
+    {
+        "role": "system",
+        "content": [{"type": "text", "text": "You are a visual affordance detection assistant. Given a first-person view image, identify what actions can be performed on the visible objects."}],
+    },
+    {
+        "role": "user",
+        "content": [
+            {"type": "image", "image": "path/to/your/image.jpg"},
+            {"type": "text", "text": "What affordances are present in this scene?"},
+        ],
+    },
+]
+
+# Generate
+input_text = tokenizer.apply_chat_template(
+    messages, tokenize=False, add_generation_prompt=True
+)
+image_inputs, _ = process_vision_info(messages)
+inputs = tokenizer(
+    text=[input_text],
+    images=image_inputs,
+    return_tensors="pt",
+    padding=True,
+).to(model.device)
+
+output_ids = model.generate(**inputs, max_new_tokens=256)
+new_tokens = output_ids[0][inputs["input_ids"].shape[1]:]
+response = tokenizer.decode(new_tokens, skip_special_tokens=True)
+print(response)
+```
+
+### Using with Transformers + PEFT (without Unsloth)
+
+```python
+import os
+os.environ["TORCHDYNAMO_DISABLE"] = "1"
+
+import torch
+from transformers import AutoProcessor, Qwen3VLForConditionalGeneration
+from peft import PeftModel
+from qwen_vl_utils import process_vision_info
+
+# Load base model in 4-bit
+from transformers import BitsAndBytesConfig
+
+bnb_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_compute_dtype=torch.bfloat16,
+)
+
+base_model = Qwen3VLForConditionalGeneration.from_pretrained(
+    "Qwen/Qwen3-VL-2B-Instruct",
+    quantization_config=bnb_config,
+    device_map="auto",
+)
+
+# Load LoRA adapter
+model = PeftModel.from_pretrained(base_model, "Kavin60606/qwen3-vl-2b-affordance-finetuned")
+processor = AutoProcessor.from_pretrained("Qwen/Qwen3-VL-2B-Instruct")
+
+# Prepare and run inference
+messages = [
+    {
+        "role": "user",
+        "content": [
+            {"type": "image", "image": "path/to/image.jpg"},
+            {"type": "text", "text": "What affordances are present in this scene?"},
+        ],
+    },
+]
+
+text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+image_inputs, _ = process_vision_info(messages)
+inputs = processor(text=[text], images=image_inputs, return_tensors="pt", padding=True).to(model.device)
+
+output = model.generate(**inputs, max_new_tokens=256)
+print(processor.decode(output[0], skip_special_tokens=True))
+```
+
+## Repository Structure
+
+```
+.
+├── adapter_config.json              # LoRA adapter configuration
+├── adapter_model.safetensors        # Fine-tuned LoRA weights (~35MB)
+├── tokenizer.json                   # Tokenizer files
+├── tokenizer_config.json
+├── preprocessor_config.json
+├── chat_template.jinja
+├── trainer_state.json               # Full training log history
+│
+├── checkpoints/
+│   ├── checkpoint-2000/             # Checkpoint at step 2000
+│   └── checkpoint-2145/             # Final checkpoint (step 2145)
+│
+├── dataset/
+│   ├── annotations.jsonl            # Original annotations (3,177 entries)
+│   ├── train_msswift.jsonl          # Training data - MS-SWIFT format (2,859)
+│   ├── val_msswift.jsonl            # Validation data - MS-SWIFT format (318)
+│   ├── train_llamafactory.json      # Training data - LLaMA-Factory format
+│   ├── val_llamafactory.json        # Validation data - LLaMA-Factory format
+│   └── dataset_info.json            # LLaMA-Factory dataset registry
+│
+├── results/
+│   ├── training_curves.png          # Training loss, grad norm, LR curves
+│   ├── loss_by_epoch.png            # Loss over epochs
+│   ├── inference_comparison.png     # Visual GT vs prediction comparison
+│   └── inference_results.json       # Full inference results (20 samples)
+│
+├── train.py                         # Training script (Unsloth + TRL)
+├── inference.py                     # Inference script
+├── analyze_and_infer.py             # Analysis + inference + visualization
+└── prepare_finetune_data.py         # Dataset preparation script
+```
+
+## Training Configuration
+
+```python
+# Key hyperparameters
+MODEL = "unsloth/Qwen3-VL-2B-Instruct"
+QUANTIZATION = "4-bit NF4"
+LORA_RANK = 16
+LORA_ALPHA = 32
+BATCH_SIZE = 1
+GRADIENT_ACCUMULATION = 4  # effective batch = 4
+LEARNING_RATE = 2e-4
+LR_SCHEDULER = "cosine"
+WARMUP_RATIO = 0.1
+EPOCHS = 3
+MAX_SEQ_LENGTH = 2048
+OPTIMIZER = "adamw_8bit"
+PRECISION = "bfloat16"
+GRADIENT_CHECKPOINTING = True
+VISION_ENCODER = "frozen"  # ViT frozen to save VRAM
+```
+
+## Reproduce Training
+
+```bash
+# 1. Clone and setup
+git clone https://huggingface.co/Kavin60606/qwen3-vl-2b-affordance-finetuned
+cd qwen3-vl-2b-affordance-finetuned
+
+# 2. Create environment
+python -m venv finetune_env
+# Linux/Mac:
+source finetune_env/bin/activate
+# Windows:
+.\finetune_env\Scripts\Activate.ps1
+
+# 3. Install dependencies
+pip install torch==2.8.0 torchvision==0.23.0 --index-url https://download.pytorch.org/whl/cu126
+pip install unsloth peft>=0.14.0 bitsandbytes>=0.49.1 accelerate>=1.2.0
+pip install trl>=0.17.0 transformers>=4.57.1 datasets qwen-vl-utils Pillow matplotlib
+
+# 4. Download the dataset images
+pip install huggingface_hub
+hf download sanskxr02/fva_affordance_v1 --repo-type dataset --local-dir ./dataset_images
+
+# 5. Run training
+# Windows:
+$env:PYTORCH_CUDA_ALLOC_CONF="expandable_segments:True"
+# Linux:
+export PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True
+
+python train.py
+```
+
+## Limitations
+
+- Trained on only **2 task scenarios** (coffee making + wine bottling) — may not generalize to other domains
+- Only **3,177 training samples** — limited diversity
+- Original annotations generated by `qwen2-vl-2b-int4` — may contain noise
+- Affordance labels are relatively short (avg 2.2 words)
+- Vision encoder was frozen during training to fit in 8GB VRAM
+
+## Citation
+
+If you use this model, please cite:
+
+```bibtex
+@misc{qwen3vl-affordance-2026,
+  title={Qwen3-VL-2B Fine-Tuned for Visual Affordance Detection},
+  author={Kavin60606},
+  year={2026},
+  publisher={HuggingFace},
+  url={https://huggingface.co/Kavin60606/qwen3-vl-2b-affordance-finetuned}
+}
+```

adapter_config.json

@@ -0,0 +1,50 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Qwen3VLForConditionalGeneration",
+    "parent_library": "transformers.models.qwen3_vl.modeling_qwen3_vl",
+    "unsloth_fixed": true
+  },
+  "base_model_name_or_path": "unsloth/qwen3-vl-2b-instruct-unsloth-bnb-4bit",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "o_proj",
+    "down_proj",
+    "q_proj",
+    "gate_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd296c2718a6c1b83cc3b19cc8ccf38f2f3b46f5336d1866a4a0f99f3173afae
+size 69788264

added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

analyze_and_infer.py

@@ -0,0 +1,310 @@
+"""
+Analyze training results + run inference on validation data.
+Generates graphs and comparison results.
+"""
+
+import os
+import json
+
+os.environ["TORCHDYNAMO_DISABLE"] = "1"
+
+import torch
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+import numpy as np
+from PIL import Image
+from unsloth import FastVisionModel
+
+# ============================================================
+# PATHS
+# ============================================================
+OUTPUT_DIR = "c:/Fidelity_dynamics/qwen_affordance_exp/output"
+DATASET_DIR = "c:/Fidelity_dynamics/qwen_affordance_exp/dataset"
+ADAPTER_PATH = os.path.join(OUTPUT_DIR, "lora_adapter")
+VAL_FILE = os.path.join(DATASET_DIR, "val_msswift.jsonl")
+RESULTS_DIR = "c:/Fidelity_dynamics/qwen_affordance_exp/results"
+os.makedirs(RESULTS_DIR, exist_ok=True)
+
+# ============================================================
+# PART 1: TRAINING LOSS ANALYSIS & GRAPHS
+# ============================================================
+print("=" * 60)
+print("PART 1: Training Analysis")
+print("=" * 60)
+
+# Load trainer state
+with open(os.path.join(OUTPUT_DIR, "checkpoint-2145", "trainer_state.json"), "r") as f:
+    state = json.load(f)
+
+log_history = state["log_history"]
+
+# Extract training metrics
+train_steps = []
+train_losses = []
+train_grad_norms = []
+train_lr = []
+eval_steps = []
+eval_losses = []
+
+for entry in log_history:
+    if "loss" in entry and "eval_loss" not in entry:
+        train_steps.append(entry["step"])
+        train_losses.append(entry["loss"])
+        train_grad_norms.append(entry.get("grad_norm", 0))
+        train_lr.append(entry.get("learning_rate", 0))
+    if "eval_loss" in entry:
+        eval_steps.append(entry["step"])
+        eval_losses.append(entry["eval_loss"])
+
+# Print summary
+print(f"Total training steps: {train_steps[-1]}")
+print(f"Initial loss: {train_losses[0]:.4f}")
+print(f"Final loss: {train_losses[-1]:.4f}")
+print(f"Best train loss: {min(train_losses):.4f} (step {train_steps[train_losses.index(min(train_losses))]})")
+if eval_losses:
+    print(f"Best eval loss: {min(eval_losses):.4f} (step {eval_steps[eval_losses.index(min(eval_losses))]})")
+    print(f"Final eval loss: {eval_losses[-1]:.4f}")
+
+# --- FIGURE 1: Training & Eval Loss ---
+fig, axes = plt.subplots(2, 2, figsize=(14, 10))
+fig.suptitle("Qwen3-VL-2B Affordance Fine-Tuning Results", fontsize=16, fontweight="bold")
+
+# Plot 1: Training Loss
+ax1 = axes[0, 0]
+ax1.plot(train_steps, train_losses, color="#2196F3", linewidth=0.8, alpha=0.6, label="Train Loss")
+# Smooth curve
+window = 20
+if len(train_losses) > window:
+    smooth_losses = np.convolve(train_losses, np.ones(window)/window, mode="valid")
+    smooth_steps = train_steps[window-1:]
+    ax1.plot(smooth_steps, smooth_losses, color="#0D47A1", linewidth=2, label=f"Smoothed (w={window})")
+if eval_losses:
+    ax1.plot(eval_steps, eval_losses, "ro-", markersize=5, linewidth=2, label="Eval Loss")
+ax1.set_xlabel("Steps")
+ax1.set_ylabel("Loss")
+ax1.set_title("Training & Evaluation Loss")
+ax1.legend()
+ax1.grid(True, alpha=0.3)
+ax1.set_ylim(bottom=0)
+
+# Plot 2: Loss (log scale for early training)
+ax2 = axes[0, 1]
+ax2.plot(train_steps, train_losses, color="#4CAF50", linewidth=0.8, alpha=0.6)
+if len(train_losses) > window:
+    ax2.plot(smooth_steps, smooth_losses, color="#1B5E20", linewidth=2)
+ax2.set_xlabel("Steps")
+ax2.set_ylabel("Loss (log scale)")
+ax2.set_title("Training Loss (Log Scale)")
+ax2.set_yscale("log")
+ax2.grid(True, alpha=0.3)
+
+# Plot 3: Gradient Norm
+ax3 = axes[1, 0]
+ax3.plot(train_steps, train_grad_norms, color="#FF9800", linewidth=0.8, alpha=0.6, label="Grad Norm")
+if len(train_grad_norms) > window:
+    smooth_grads = np.convolve(train_grad_norms, np.ones(window)/window, mode="valid")
+    ax3.plot(smooth_steps, smooth_grads, color="#E65100", linewidth=2, label=f"Smoothed (w={window})")
+ax3.set_xlabel("Steps")
+ax3.set_ylabel("Gradient Norm")
+ax3.set_title("Gradient Norm Over Training")
+ax3.legend()
+ax3.grid(True, alpha=0.3)
+
+# Plot 4: Learning Rate Schedule
+ax4 = axes[1, 1]
+ax4.plot(train_steps, [lr * 1e4 for lr in train_lr], color="#9C27B0", linewidth=2)
+ax4.set_xlabel("Steps")
+ax4.set_ylabel("Learning Rate (×1e-4)")
+ax4.set_title("Learning Rate Schedule (Cosine)")
+ax4.grid(True, alpha=0.3)
+
+plt.tight_layout()
+plt.savefig(os.path.join(RESULTS_DIR, "training_curves.png"), dpi=150, bbox_inches="tight")
+print(f"\nSaved: {os.path.join(RESULTS_DIR, 'training_curves.png')}")
+plt.close()
+
+# --- FIGURE 2: Loss by Epoch ---
+fig2, ax = plt.subplots(figsize=(10, 5))
+epochs = [entry.get("epoch", 0) for entry in log_history if "loss" in entry and "eval_loss" not in entry]
+ax.plot(epochs, train_losses, color="#2196F3", linewidth=0.8, alpha=0.5)
+if len(train_losses) > window:
+    smooth_epochs = epochs[window-1:]
+    ax.plot(smooth_epochs, smooth_losses, color="#0D47A1", linewidth=2, label="Smoothed Loss")
+ax.axvline(x=1.0, color="gray", linestyle="--", alpha=0.5, label="Epoch boundary")
+ax.axvline(x=2.0, color="gray", linestyle="--", alpha=0.5)
+ax.set_xlabel("Epoch")
+ax.set_ylabel("Loss")
+ax.set_title("Training Loss by Epoch")
+ax.legend()
+ax.grid(True, alpha=0.3)
+plt.tight_layout()
+plt.savefig(os.path.join(RESULTS_DIR, "loss_by_epoch.png"), dpi=150, bbox_inches="tight")
+print(f"Saved: {os.path.join(RESULTS_DIR, 'loss_by_epoch.png')}")
+plt.close()
+
+
+# ============================================================
+# PART 2: INFERENCE ON VALIDATION DATA
+# ============================================================
+print("\n" + "=" * 60)
+print("PART 2: Running Inference on Validation Data")
+print("=" * 60)
+
+# Load model with adapter
+print("Loading fine-tuned model...")
+model, tokenizer = FastVisionModel.from_pretrained(
+    ADAPTER_PATH,
+    load_in_4bit=True,
+    max_seq_length=2048,
+)
+FastVisionModel.for_inference(model)
+print("Model loaded!")
+
+# Load validation data
+val_data = []
+with open(VAL_FILE, "r", encoding="utf-8") as f:
+    for line in f:
+        val_data.append(json.loads(line.strip()))
+
+print(f"Validation samples: {len(val_data)}")
+
+# Run inference on 20 samples
+num_samples = min(20, len(val_data))
+results = []
+
+for i in range(num_samples):
+    sample = val_data[i]
+    img_path = sample["images"][0]
+    ground_truth = sample["messages"][2]["content"]
+    question = sample["messages"][1]["content"].replace("<image>\n", "").replace("<image>", "")
+
+    # Load image
+    try:
+        image = Image.open(img_path).convert("RGB")
+        image = image.resize((384, 216), Image.LANCZOS)
+    except Exception as e:
+        print(f"  Skipping {img_path}: {e}")
+        continue
+
+    # Build chat messages with image path for processor
+    messages = [
+        {
+            "role": "system",
+            "content": [{"type": "text", "text": sample["messages"][0]["content"]}],
+        },
+        {
+            "role": "user",
+            "content": [
+                {"type": "image", "image": img_path},
+                {"type": "text", "text": question},
+            ],
+        },
+    ]
+
+    # Use processor to handle both text and images
+    from qwen_vl_utils import process_vision_info
+    input_text = tokenizer.apply_chat_template(
+        messages, tokenize=False, add_generation_prompt=True
+    )
+    image_inputs, _ = process_vision_info(messages)
+    inputs = tokenizer(
+        text=[input_text],
+        images=image_inputs,
+        return_tensors="pt",
+        padding=True,
+    ).to(model.device)
+
+    # Generate
+    with torch.no_grad():
+        output_ids = model.generate(
+            **inputs,
+            max_new_tokens=256,
+            temperature=0.7,
+            do_sample=True,
+        )
+
+    new_tokens = output_ids[0][inputs["input_ids"].shape[1]:]
+    prediction = tokenizer.decode(new_tokens, skip_special_tokens=True)
+
+    results.append({
+        "image": os.path.basename(img_path),
+        "question": question,
+        "ground_truth": ground_truth,
+        "prediction": prediction,
+    })
+
+    print(f"\n--- Sample {i+1}/{num_samples} ---")
+    print(f"  Image: {os.path.basename(img_path)}")
+    print(f"  GT:   {ground_truth[:120]}...")
+    print(f"  Pred: {prediction[:120]}...")
+
+# Save results
+results_path = os.path.join(RESULTS_DIR, "inference_results.json")
+with open(results_path, "w", encoding="utf-8") as f:
+    json.dump(results, f, indent=2, ensure_ascii=False)
+print(f"\nSaved: {results_path}")
+
+# ============================================================
+# PART 3: RESULTS COMPARISON FIGURE
+# ============================================================
+print("\n" + "=" * 60)
+print("PART 3: Creating Comparison Visuals")
+print("=" * 60)
+
+# Create a comparison figure showing images with GT vs Prediction
+num_vis = min(6, len(results))
+fig3, axes = plt.subplots(2, 3, figsize=(18, 10))
+fig3.suptitle("Inference Results: Ground Truth vs Prediction", fontsize=14, fontweight="bold")
+
+for idx in range(num_vis):
+    r = results[idx]
+    ax = axes[idx // 3][idx % 3]
+
+    # Load image
+    img_path_full = val_data[idx]["images"][0]
+    try:
+        img = Image.open(img_path_full).convert("RGB")
+        ax.imshow(img)
+    except:
+        ax.text(0.5, 0.5, "Image not found", ha="center", va="center")
+
+    # Extract just the affordance from GT and prediction
+    gt_short = r["ground_truth"][:80]
+    pred_short = r["prediction"][:80]
+
+    ax.set_title(f"GT: {gt_short}...\nPred: {pred_short}...", fontsize=7, wrap=True)
+    ax.axis("off")
+
+plt.tight_layout()
+plt.savefig(os.path.join(RESULTS_DIR, "inference_comparison.png"), dpi=150, bbox_inches="tight")
+print(f"Saved: {os.path.join(RESULTS_DIR, 'inference_comparison.png')}")
+plt.close()
+
+# ============================================================
+# PART 4: SUMMARY STATS
+# ============================================================
+print("\n" + "=" * 60)
+print("TRAINING SUMMARY")
+print("=" * 60)
+
+runtime = state["log_history"][-1].get("train_runtime", 0)
+print(f"  Model:             Qwen3-VL-2B-Instruct")
+print(f"  Method:            QLoRA 4-bit (rank=16)")
+print(f"  Trainable params:  17.4M / 1.52B (1.15%)")
+print(f"  Training samples:  2,859")
+print(f"  Validation samples: 318")
+print(f"  Epochs:            3")
+print(f"  Total steps:       2,145")
+print(f"  Training time:     {runtime/60:.1f} minutes")
+print(f"  Initial loss:      {train_losses[0]:.4f}")
+print(f"  Final loss:        {train_losses[-1]:.4f}")
+print(f"  Loss reduction:    {((train_losses[0] - train_losses[-1]) / train_losses[0] * 100):.1f}%")
+if eval_losses:
+    print(f"  Best eval loss:    {min(eval_losses):.4f}")
+    print(f"  Final eval loss:   {eval_losses[-1]:.4f}")
+print(f"  GPU:               RTX 3060 Ti (8GB)")
+print(f"  Adapter size:      saved to output/lora_adapter/")
+print(f"\n  Results saved to:  {RESULTS_DIR}/")
+
+print("\nDone!")

chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

checkpoints/checkpoint-2000/README.md

@@ -0,0 +1,210 @@
+---
+base_model: unsloth/qwen3-vl-2b-instruct-unsloth-bnb-4bit
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:unsloth/qwen3-vl-2b-instruct-unsloth-bnb-4bit
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

checkpoints/checkpoint-2000/adapter_config.json

@@ -0,0 +1,50 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Qwen3VLForConditionalGeneration",
+    "parent_library": "transformers.models.qwen3_vl.modeling_qwen3_vl",
+    "unsloth_fixed": true
+  },
+  "base_model_name_or_path": "unsloth/qwen3-vl-2b-instruct-unsloth-bnb-4bit",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "o_proj",
+    "down_proj",
+    "q_proj",
+    "gate_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

checkpoints/checkpoint-2000/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8766ab2dc24b942396ca5c31190ed853a92763cf5e615f2eb34512d2c94f0ade
+size 69788264

checkpoints/checkpoint-2000/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

checkpoints/checkpoint-2000/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

checkpoints/checkpoint-2000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:381cad2f239dc4e87b6dd3839e181cdfd02bc786c04d914183dd8d037426efa0
+size 35869445

checkpoints/checkpoint-2000/preprocessor_config.json

@@ -0,0 +1,39 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": true,
+  "device": null,
+  "disable_grouping": null,
+  "do_center_crop": null,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_pad": null,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "Qwen2VLImageProcessorFast",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "input_data_format": null,
+  "max_pixels": null,
+  "merge_size": 2,
+  "min_pixels": null,
+  "pad_size": null,
+  "patch_size": 16,
+  "processor_class": "Qwen3VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "return_tensors": null,
+  "size": {
+    "longest_edge": 16777216,
+    "shortest_edge": 65536
+  },
+  "temporal_patch_size": 2
+}

checkpoints/checkpoint-2000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89f2e7b9901fa4b72af37f5845fbfc651b4a741e105cc6dfc15f2793ce6c6f67
+size 14645

checkpoints/checkpoint-2000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c8a23b366aae2c2a3fd79efe1885b8532ff0e6c1470f4600f7d9e1a767537f3
+size 1465

checkpoints/checkpoint-2000/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|vision_pad|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoints/checkpoint-2000/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af560c5e5807192776a9f72a0ea33c9331816d9c59af61e02bdf766f8a2a97fb
+size 11422922

checkpoints/checkpoint-2000/tokenizer_config.json

@@ -0,0 +1,241 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 262144,
+  "pad_token": "<|vision_pad|>",
+  "padding_side": "right",
+  "processor_class": "Qwen3VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

checkpoints/checkpoint-2000/trainer_state.json

@@ -0,0 +1,1514 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.7974816369359914,
+  "eval_steps": 200,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013990905911157748,
+      "grad_norm": 11.852328300476074,
+      "learning_rate": 8.372093023255815e-06,
+      "loss": 3.4227,
+      "step": 10
+    },
+    {
+      "epoch": 0.027981811822315496,
+      "grad_norm": 5.509212017059326,
+      "learning_rate": 1.7674418604651163e-05,
+      "loss": 3.1068,
+      "step": 20
+    },
+    {
+      "epoch": 0.04197271773347324,
+      "grad_norm": 3.556685209274292,
+      "learning_rate": 2.6976744186046517e-05,
+      "loss": 2.3908,
+      "step": 30
+    },
+    {
+      "epoch": 0.05596362364463099,
+      "grad_norm": 4.118677616119385,
+      "learning_rate": 3.627906976744187e-05,
+      "loss": 1.5286,
+      "step": 40
+    },
+    {
+      "epoch": 0.06995452955578874,
+      "grad_norm": 2.5455708503723145,
+      "learning_rate": 4.5581395348837214e-05,
+      "loss": 0.7724,
+      "step": 50
+    },
+    {
+      "epoch": 0.08394543546694648,
+      "grad_norm": 1.4716579914093018,
+      "learning_rate": 5.488372093023256e-05,
+      "loss": 0.3533,
+      "step": 60
+    },
+    {
+      "epoch": 0.09793634137810424,
+      "grad_norm": 1.154810905456543,
+      "learning_rate": 6.418604651162791e-05,
+      "loss": 0.2292,
+      "step": 70
+    },
+    {
+      "epoch": 0.11192724728926198,
+      "grad_norm": 0.8629467487335205,
+      "learning_rate": 7.348837209302326e-05,
+      "loss": 0.1772,
+      "step": 80
+    },
+    {
+      "epoch": 0.1259181532004197,
+      "grad_norm": 0.7331057190895081,
+      "learning_rate": 8.27906976744186e-05,
+      "loss": 0.1643,
+      "step": 90
+    },
+    {
+      "epoch": 0.13990905911157747,
+      "grad_norm": 0.5298661589622498,
+      "learning_rate": 9.209302325581396e-05,
+      "loss": 0.1659,
+      "step": 100
+    },
+    {
+      "epoch": 0.15389996502273523,
+      "grad_norm": 0.6423642635345459,
+      "learning_rate": 0.00010139534883720931,
+      "loss": 0.1569,
+      "step": 110
+    },
+    {
+      "epoch": 0.16789087093389296,
+      "grad_norm": 0.7554107904434204,
+      "learning_rate": 0.00011069767441860466,
+      "loss": 0.1335,
+      "step": 120
+    },
+    {
+      "epoch": 0.18188177684505072,
+      "grad_norm": 0.6365725994110107,
+      "learning_rate": 0.00012,
+      "loss": 0.1346,
+      "step": 130
+    },
+    {
+      "epoch": 0.19587268275620848,
+      "grad_norm": 0.577471137046814,
+      "learning_rate": 0.00012930232558139534,
+      "loss": 0.1559,
+      "step": 140
+    },
+    {
+      "epoch": 0.2098635886673662,
+      "grad_norm": 0.6245256066322327,
+      "learning_rate": 0.00013860465116279072,
+      "loss": 0.1678,
+      "step": 150
+    },
+    {
+      "epoch": 0.22385449457852397,
+      "grad_norm": 0.5085740089416504,
+      "learning_rate": 0.00014790697674418605,
+      "loss": 0.162,
+      "step": 160
+    },
+    {
+      "epoch": 0.2378454004896817,
+      "grad_norm": 0.46424826979637146,
+      "learning_rate": 0.0001572093023255814,
+      "loss": 0.1328,
+      "step": 170
+    },
+    {
+      "epoch": 0.2518363064008394,
+      "grad_norm": 0.4451293647289276,
+      "learning_rate": 0.00016651162790697673,
+      "loss": 0.1528,
+      "step": 180
+    },
+    {
+      "epoch": 0.2658272123119972,
+      "grad_norm": 0.6359198689460754,
+      "learning_rate": 0.0001758139534883721,
+      "loss": 0.1334,
+      "step": 190
+    },
+    {
+      "epoch": 0.27981811822315494,
+      "grad_norm": 0.406089723110199,
+      "learning_rate": 0.00018511627906976744,
+      "loss": 0.1414,
+      "step": 200
+    },
+    {
+      "epoch": 0.27981811822315494,
+      "eval_loss": 0.1422451138496399,
+      "eval_runtime": 21.5015,
+      "eval_samples_per_second": 14.79,
+      "eval_steps_per_second": 3.721,
+      "step": 200
+    },
+    {
+      "epoch": 0.2938090241343127,
+      "grad_norm": 0.4091590642929077,
+      "learning_rate": 0.0001944186046511628,
+      "loss": 0.1526,
+      "step": 210
+    },
+    {
+      "epoch": 0.30779993004547046,
+      "grad_norm": 0.325090229511261,
+      "learning_rate": 0.000199997880304001,
+      "loss": 0.1173,
+      "step": 220
+    },
+    {
+      "epoch": 0.3217908359566282,
+      "grad_norm": 0.3782133460044861,
+      "learning_rate": 0.0001999740347560115,
+      "loss": 0.1445,
+      "step": 230
+    },
+    {
+      "epoch": 0.3357817418677859,
+      "grad_norm": 0.4928087592124939,
+      "learning_rate": 0.00019992370037914218,
+      "loss": 0.1363,
+      "step": 240
+    },
+    {
+      "epoch": 0.3497726477789437,
+      "grad_norm": 0.31038427352905273,
+      "learning_rate": 0.00019984689050984278,
+      "loss": 0.1296,
+      "step": 250
+    },
+    {
+      "epoch": 0.36376355369010144,
+      "grad_norm": 0.4541214108467102,
+      "learning_rate": 0.00019974362549943247,
+      "loss": 0.1207,
+      "step": 260
+    },
+    {
+      "epoch": 0.3777544596012592,
+      "grad_norm": 0.46741530299186707,
+      "learning_rate": 0.00019961393270870738,
+      "loss": 0.1386,
+      "step": 270
+    },
+    {
+      "epoch": 0.39174536551241695,
+      "grad_norm": 0.3349354863166809,
+      "learning_rate": 0.00019945784650069143,
+      "loss": 0.1533,
+      "step": 280
+    },
+    {
+      "epoch": 0.40573627142357466,
+      "grad_norm": 0.40393367409706116,
+      "learning_rate": 0.00019927540823153135,
+      "loss": 0.135,
+      "step": 290
+    },
+    {
+      "epoch": 0.4197271773347324,
+      "grad_norm": 0.34774526953697205,
+      "learning_rate": 0.00019906666623953925,
+      "loss": 0.1213,
+      "step": 300
+    },
+    {
+      "epoch": 0.4337180832458902,
+      "grad_norm": 0.35436466336250305,
+      "learning_rate": 0.00019883167583238494,
+      "loss": 0.1244,
+      "step": 310
+    },
+    {
+      "epoch": 0.44770898915704793,
+      "grad_norm": 0.3164362609386444,
+      "learning_rate": 0.00019857049927244185,
+      "loss": 0.1243,
+      "step": 320
+    },
+    {
+      "epoch": 0.4616998950682057,
+      "grad_norm": 0.3490104675292969,
+      "learning_rate": 0.0001982832057602901,
+      "loss": 0.1427,
+      "step": 330
+    },
+    {
+      "epoch": 0.4756908009793634,
+      "grad_norm": 0.4742378294467926,
+      "learning_rate": 0.00019796987141638134,
+      "loss": 0.1252,
+      "step": 340
+    },
+    {
+      "epoch": 0.48968170689052115,
+      "grad_norm": 0.33545687794685364,
+      "learning_rate": 0.00019763057926087022,
+      "loss": 0.1355,
+      "step": 350
+    },
+    {
+      "epoch": 0.5036726128016789,
+      "grad_norm": 0.4260849058628082,
+      "learning_rate": 0.00019726541919161733,
+      "loss": 0.1125,
+      "step": 360
+    },
+    {
+      "epoch": 0.5176635187128367,
+      "grad_norm": 0.4119410514831543,
+      "learning_rate": 0.00019687448796037045,
+      "loss": 0.122,
+      "step": 370
+    },
+    {
+      "epoch": 0.5316544246239944,
+      "grad_norm": 0.28887462615966797,
+      "learning_rate": 0.0001964578891471292,
+      "loss": 0.129,
+      "step": 380
+    },
+    {
+      "epoch": 0.5456453305351522,
+      "grad_norm": 0.3469299077987671,
+      "learning_rate": 0.00019601573313270102,
+      "loss": 0.1211,
+      "step": 390
+    },
+    {
+      "epoch": 0.5596362364463099,
+      "grad_norm": 0.2652361989021301,
+      "learning_rate": 0.00019554813706945486,
+      "loss": 0.1217,
+      "step": 400
+    },
+    {
+      "epoch": 0.5596362364463099,
+      "eval_loss": 0.12272346764802933,
+      "eval_runtime": 21.4783,
+      "eval_samples_per_second": 14.806,
+      "eval_steps_per_second": 3.725,
+      "step": 400
+    },
+    {
+      "epoch": 0.5736271423574677,
+      "grad_norm": 0.4036720395088196,
+      "learning_rate": 0.0001950552248502808,
+      "loss": 0.1341,
+      "step": 410
+    },
+    {
+      "epoch": 0.5876180482686254,
+      "grad_norm": 0.4273042380809784,
+      "learning_rate": 0.00019453712707576392,
+      "loss": 0.1093,
+      "step": 420
+    },
+    {
+      "epoch": 0.6016089541797831,
+      "grad_norm": 0.3455790877342224,
+      "learning_rate": 0.00019399398101958056,
+      "loss": 0.1133,
+      "step": 430
+    },
+    {
+      "epoch": 0.6155998600909409,
+      "grad_norm": 0.2522144019603729,
+      "learning_rate": 0.0001934259305921268,
+      "loss": 0.1221,
+      "step": 440
+    },
+    {
+      "epoch": 0.6295907660020986,
+      "grad_norm": 0.21513080596923828,
+      "learning_rate": 0.00019283312630238836,
+      "loss": 0.1112,
+      "step": 450
+    },
+    {
+      "epoch": 0.6435816719132564,
+      "grad_norm": 0.2985060214996338,
+      "learning_rate": 0.00019221572521806212,
+      "loss": 0.1072,
+      "step": 460
+    },
+    {
+      "epoch": 0.6575725778244141,
+      "grad_norm": 0.23942740261554718,
+      "learning_rate": 0.00019157389092394004,
+      "loss": 0.113,
+      "step": 470
+    },
+    {
+      "epoch": 0.6715634837355718,
+      "grad_norm": 0.35044464468955994,
+      "learning_rate": 0.00019090779347856606,
+      "loss": 0.1186,
+      "step": 480
+    },
+    {
+      "epoch": 0.6855543896467297,
+      "grad_norm": 0.25626683235168457,
+      "learning_rate": 0.00019021760936917804,
+      "loss": 0.1156,
+      "step": 490
+    },
+    {
+      "epoch": 0.6995452955578874,
+      "grad_norm": 0.3051966428756714,
+      "learning_rate": 0.00018950352146494606,
+      "loss": 0.1166,
+      "step": 500
+    },
+    {
+      "epoch": 0.7135362014690452,
+      "grad_norm": 0.3047504127025604,
+      "learning_rate": 0.00018876571896852005,
+      "loss": 0.1155,
+      "step": 510
+    },
+    {
+      "epoch": 0.7275271073802029,
+      "grad_norm": 0.2766532301902771,
+      "learning_rate": 0.00018800439736589924,
+      "loss": 0.1144,
+      "step": 520
+    },
+    {
+      "epoch": 0.7415180132913606,
+      "grad_norm": 0.321286141872406,
+      "learning_rate": 0.00018721975837463655,
+      "loss": 0.1032,
+      "step": 530
+    },
+    {
+      "epoch": 0.7555089192025184,
+      "grad_norm": 0.27030113339424133,
+      "learning_rate": 0.00018641200989039236,
+      "loss": 0.1201,
+      "step": 540
+    },
+    {
+      "epoch": 0.7694998251136761,
+      "grad_norm": 0.3184170424938202,
+      "learning_rate": 0.00018558136593185086,
+      "loss": 0.1234,
+      "step": 550
+    },
+    {
+      "epoch": 0.7834907310248339,
+      "grad_norm": 0.2622183859348297,
+      "learning_rate": 0.00018472804658401453,
+      "loss": 0.1015,
+      "step": 560
+    },
+    {
+      "epoch": 0.7974816369359916,
+      "grad_norm": 0.3095245659351349,
+      "learning_rate": 0.00018385227793989075,
+      "loss": 0.122,
+      "step": 570
+    },
+    {
+      "epoch": 0.8114725428471493,
+      "grad_norm": 0.4251258969306946,
+      "learning_rate": 0.00018295429204058712,
+      "loss": 0.1167,
+      "step": 580
+    },
+    {
+      "epoch": 0.8254634487583071,
+      "grad_norm": 0.31186479330062866,
+      "learning_rate": 0.00018203432681383034,
+      "loss": 0.1114,
+      "step": 590
+    },
+    {
+      "epoch": 0.8394543546694648,
+      "grad_norm": 0.2990236282348633,
+      "learning_rate": 0.00018109262601092576,
+      "loss": 0.1103,
+      "step": 600
+    },
+    {
+      "epoch": 0.8394543546694648,
+      "eval_loss": 0.11650671809911728,
+      "eval_runtime": 21.574,
+      "eval_samples_per_second": 14.74,
+      "eval_steps_per_second": 3.708,
+      "step": 600
+    },
+    {
+      "epoch": 0.8534452605806226,
+      "grad_norm": 0.3199589252471924,
+      "learning_rate": 0.00018012943914217356,
+      "loss": 0.1098,
+      "step": 610
+    },
+    {
+      "epoch": 0.8674361664917803,
+      "grad_norm": 0.2671171724796295,
+      "learning_rate": 0.00017914502141075928,
+      "loss": 0.0959,
+      "step": 620
+    },
+    {
+      "epoch": 0.881427072402938,
+      "grad_norm": 0.371975302696228,
+      "learning_rate": 0.00017813963364513607,
+      "loss": 0.1089,
+      "step": 630
+    },
+    {
+      "epoch": 0.8954179783140959,
+      "grad_norm": 0.28194791078567505,
+      "learning_rate": 0.00017711354222991606,
+      "loss": 0.1009,
+      "step": 640
+    },
+    {
+      "epoch": 0.9094088842252536,
+      "grad_norm": 0.24068565666675568,
+      "learning_rate": 0.0001760670190352903,
+      "loss": 0.1195,
+      "step": 650
+    },
+    {
+      "epoch": 0.9233997901364114,
+      "grad_norm": 0.35747045278549194,
+      "learning_rate": 0.00017500034134499423,
+      "loss": 0.1237,
+      "step": 660
+    },
+    {
+      "epoch": 0.9373906960475691,
+      "grad_norm": 0.30066290497779846,
+      "learning_rate": 0.00017391379178284002,
+      "loss": 0.1231,
+      "step": 670
+    },
+    {
+      "epoch": 0.9513816019587268,
+      "grad_norm": 0.3015068769454956,
+      "learning_rate": 0.0001728076582378329,
+      "loss": 0.1,
+      "step": 680
+    },
+    {
+      "epoch": 0.9653725078698846,
+      "grad_norm": 0.3291875422000885,
+      "learning_rate": 0.0001716822337878932,
+      "loss": 0.1256,
+      "step": 690
+    },
+    {
+      "epoch": 0.9793634137810423,
+      "grad_norm": 0.2612892985343933,
+      "learning_rate": 0.0001705378166222032,
+      "loss": 0.1114,
+      "step": 700
+    },
+    {
+      "epoch": 0.9933543196922001,
+      "grad_norm": 0.24723578989505768,
+      "learning_rate": 0.00016937470996219973,
+      "loss": 0.1021,
+      "step": 710
+    },
+    {
+      "epoch": 1.006995452955579,
+      "grad_norm": 0.21942616999149323,
+      "learning_rate": 0.00016819322198123365,
+      "loss": 0.1038,
+      "step": 720
+    },
+    {
+      "epoch": 1.0209863588667367,
+      "grad_norm": 0.2657986581325531,
+      "learning_rate": 0.00016699366572291702,
+      "loss": 0.0924,
+      "step": 730
+    },
+    {
+      "epoch": 1.0349772647778943,
+      "grad_norm": 0.3046330213546753,
+      "learning_rate": 0.00016577635901818,
+      "loss": 0.095,
+      "step": 740
+    },
+    {
+      "epoch": 1.0489681706890521,
+      "grad_norm": 0.29112890362739563,
+      "learning_rate": 0.0001645416244010593,
+      "loss": 0.0939,
+      "step": 750
+    },
+    {
+      "epoch": 1.06295907660021,
+      "grad_norm": 0.2303277999162674,
+      "learning_rate": 0.00016328978902324046,
+      "loss": 0.0936,
+      "step": 760
+    },
+    {
+      "epoch": 1.0769499825113675,
+      "grad_norm": 0.21349187195301056,
+      "learning_rate": 0.00016202118456737683,
+      "loss": 0.1021,
+      "step": 770
+    },
+    {
+      "epoch": 1.0909408884225253,
+      "grad_norm": 0.2621854245662689,
+      "learning_rate": 0.0001607361471592077,
+      "loss": 0.0998,
+      "step": 780
+    },
+    {
+      "epoch": 1.1049317943336832,
+      "grad_norm": 0.21957777440547943,
+      "learning_rate": 0.0001594350172784996,
+      "loss": 0.1091,
+      "step": 790
+    },
+    {
+      "epoch": 1.1189227002448408,
+      "grad_norm": 0.27233201265335083,
+      "learning_rate": 0.0001581181396688337,
+      "loss": 0.093,
+      "step": 800
+    },
+    {
+      "epoch": 1.1189227002448408,
+      "eval_loss": 0.1152026355266571,
+      "eval_runtime": 21.6054,
+      "eval_samples_per_second": 14.719,
+      "eval_steps_per_second": 3.703,
+      "step": 800
+    },
+    {
+      "epoch": 1.1329136061559986,
+      "grad_norm": 0.3158034086227417,
+      "learning_rate": 0.00015678586324626337,
+      "loss": 0.0907,
+      "step": 810
+    },
+    {
+      "epoch": 1.1469045120671564,
+      "grad_norm": 0.30520495772361755,
+      "learning_rate": 0.00015543854100686676,
+      "loss": 0.1057,
+      "step": 820
+    },
+    {
+      "epoch": 1.1608954179783142,
+      "grad_norm": 0.21135452389717102,
+      "learning_rate": 0.0001540765299332177,
+      "loss": 0.0929,
+      "step": 830
+    },
+    {
+      "epoch": 1.1748863238894718,
+      "grad_norm": 0.32276666164398193,
+      "learning_rate": 0.00015270019089980077,
+      "loss": 0.1085,
+      "step": 840
+    },
+    {
+      "epoch": 1.1888772298006296,
+      "grad_norm": 0.27095136046409607,
+      "learning_rate": 0.00015130988857739498,
+      "loss": 0.0984,
+      "step": 850
+    },
+    {
+      "epoch": 1.2028681357117874,
+      "grad_norm": 0.4224928915500641,
+      "learning_rate": 0.0001499059913364517,
+      "loss": 0.1117,
+      "step": 860
+    },
+    {
+      "epoch": 1.216859041622945,
+      "grad_norm": 0.24559757113456726,
+      "learning_rate": 0.00014848887114949234,
+      "loss": 0.101,
+      "step": 870
+    },
+    {
+      "epoch": 1.2308499475341028,
+      "grad_norm": 0.2625027894973755,
+      "learning_rate": 0.00014705890349255149,
+      "loss": 0.1113,
+      "step": 880
+    },
+    {
+      "epoch": 1.2448408534452606,
+      "grad_norm": 0.31155043840408325,
+      "learning_rate": 0.00014561646724569198,
+      "loss": 0.0903,
+      "step": 890
+    },
+    {
+      "epoch": 1.2588317593564184,
+      "grad_norm": 0.2001536637544632,
+      "learning_rate": 0.00014416194459261785,
+      "loss": 0.123,
+      "step": 900
+    },
+    {
+      "epoch": 1.272822665267576,
+      "grad_norm": 0.23669183254241943,
+      "learning_rate": 0.00014269572091941214,
+      "loss": 0.109,
+      "step": 910
+    },
+    {
+      "epoch": 1.2868135711787339,
+      "grad_norm": 0.2512431740760803,
+      "learning_rate": 0.00014121818471242608,
+      "loss": 0.1022,
+      "step": 920
+    },
+    {
+      "epoch": 1.3008044770898914,
+      "grad_norm": 0.26644250750541687,
+      "learning_rate": 0.00013972972745534686,
+      "loss": 0.0991,
+      "step": 930
+    },
+    {
+      "epoch": 1.3147953830010493,
+      "grad_norm": 0.24515704810619354,
+      "learning_rate": 0.0001382307435254713,
+      "loss": 0.1002,
+      "step": 940
+    },
+    {
+      "epoch": 1.328786288912207,
+      "grad_norm": 0.2648073732852936,
+      "learning_rate": 0.0001367216300892127,
+      "loss": 0.1074,
+      "step": 950
+    },
+    {
+      "epoch": 1.3427771948233649,
+      "grad_norm": 0.2987382411956787,
+      "learning_rate": 0.00013520278699686897,
+      "loss": 0.1017,
+      "step": 960
+    },
+    {
+      "epoch": 1.3567681007345227,
+      "grad_norm": 0.22436612844467163,
+      "learning_rate": 0.00013367461667667923,
+      "loss": 0.0923,
+      "step": 970
+    },
+    {
+      "epoch": 1.3707590066456803,
+      "grad_norm": 0.6144154667854309,
+      "learning_rate": 0.00013213752402819773,
+      "loss": 0.1021,
+      "step": 980
+    },
+    {
+      "epoch": 1.384749912556838,
+      "grad_norm": 0.21033185720443726,
+      "learning_rate": 0.00013059191631501288,
+      "loss": 0.1136,
+      "step": 990
+    },
+    {
+      "epoch": 1.3987408184679957,
+      "grad_norm": 0.3320418894290924,
+      "learning_rate": 0.00012903820305683973,
+      "loss": 0.1065,
+      "step": 1000
+    },
+    {
+      "epoch": 1.3987408184679957,
+      "eval_loss": 0.10766011476516724,
+      "eval_runtime": 21.6633,
+      "eval_samples_per_second": 14.679,
+      "eval_steps_per_second": 3.693,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4127317243791535,
+      "grad_norm": 0.3217187523841858,
+      "learning_rate": 0.00012747679592101498,
+      "loss": 0.1025,
+      "step": 1010
+    },
+    {
+      "epoch": 1.4267226302903113,
+      "grad_norm": 0.19164179265499115,
+      "learning_rate": 0.00012590810861342266,
+      "loss": 0.0903,
+      "step": 1020
+    },
+    {
+      "epoch": 1.4407135362014691,
+      "grad_norm": 0.30840256810188293,
+      "learning_rate": 0.00012433255676888,
+      "loss": 0.0895,
+      "step": 1030
+    },
+    {
+      "epoch": 1.4547044421126267,
+      "grad_norm": 0.25482040643692017,
+      "learning_rate": 0.00012275055784101178,
+      "loss": 0.094,
+      "step": 1040
+    },
+    {
+      "epoch": 1.4686953480237845,
+      "grad_norm": 0.22336170077323914,
+      "learning_rate": 0.00012116253099164334,
+      "loss": 0.0907,
+      "step": 1050
+    },
+    {
+      "epoch": 1.4826862539349424,
+      "grad_norm": 0.3286293148994446,
+      "learning_rate": 0.00011956889697974045,
+      "loss": 0.1058,
+      "step": 1060
+    },
+    {
+      "epoch": 1.4966771598461,
+      "grad_norm": 0.28686946630477905,
+      "learning_rate": 0.00011797007804992633,
+      "loss": 0.1009,
+      "step": 1070
+    },
+    {
+      "epoch": 1.5106680657572578,
+      "grad_norm": 0.2923101782798767,
+      "learning_rate": 0.00011636649782060496,
+      "loss": 0.0993,
+      "step": 1080
+    },
+    {
+      "epoch": 1.5246589716684156,
+      "grad_norm": 0.15968987345695496,
+      "learning_rate": 0.00011475858117172034,
+      "loss": 0.0866,
+      "step": 1090
+    },
+    {
+      "epoch": 1.5386498775795734,
+      "grad_norm": 0.2587738037109375,
+      "learning_rate": 0.00011314675413218156,
+      "loss": 0.0874,
+      "step": 1100
+    },
+    {
+      "epoch": 1.552640783490731,
+      "grad_norm": 0.2577781081199646,
+      "learning_rate": 0.00011153144376698347,
+      "loss": 0.0923,
+      "step": 1110
+    },
+    {
+      "epoch": 1.5666316894018888,
+      "grad_norm": 0.27181577682495117,
+      "learning_rate": 0.00010991307806405269,
+      "loss": 0.1007,
+      "step": 1120
+    },
+    {
+      "epoch": 1.5806225953130464,
+      "grad_norm": 0.19265908002853394,
+      "learning_rate": 0.00010829208582084915,
+      "loss": 0.0997,
+      "step": 1130
+    },
+    {
+      "epoch": 1.5946135012242042,
+      "grad_norm": 0.21604153513908386,
+      "learning_rate": 0.0001066688965307534,
+      "loss": 0.0954,
+      "step": 1140
+    },
+    {
+      "epoch": 1.608604407135362,
+      "grad_norm": 0.18990111351013184,
+      "learning_rate": 0.00010504394026926902,
+      "loss": 0.0976,
+      "step": 1150
+    },
+    {
+      "epoch": 1.6225953130465198,
+      "grad_norm": 0.28930869698524475,
+      "learning_rate": 0.00010341764758007118,
+      "loss": 0.1064,
+      "step": 1160
+    },
+    {
+      "epoch": 1.6365862189576776,
+      "grad_norm": 0.3197149634361267,
+      "learning_rate": 0.0001017904493609311,
+      "loss": 0.0999,
+      "step": 1170
+    },
+    {
+      "epoch": 1.6505771248688352,
+      "grad_norm": 0.3212111294269562,
+      "learning_rate": 0.0001001627767495465,
+      "loss": 0.092,
+      "step": 1180
+    },
+    {
+      "epoch": 1.664568030779993,
+      "grad_norm": 0.1536320447921753,
+      "learning_rate": 9.853506100930875e-05,
+      "loss": 0.0979,
+      "step": 1190
+    },
+    {
+      "epoch": 1.6785589366911506,
+      "grad_norm": 0.3106030821800232,
+      "learning_rate": 9.690773341503649e-05,
+      "loss": 0.0845,
+      "step": 1200
+    },
+    {
+      "epoch": 1.6785589366911506,
+      "eval_loss": 0.10427355021238327,
+      "eval_runtime": 22.3655,
+      "eval_samples_per_second": 14.218,
+      "eval_steps_per_second": 3.577,
+      "step": 1200
+    },
+    {
+      "epoch": 1.6925498426023085,
+      "grad_norm": 0.29129135608673096,
+      "learning_rate": 9.528122513870631e-05,
+      "loss": 0.0981,
+      "step": 1210
+    },
+    {
+      "epoch": 1.7065407485134663,
+      "grad_norm": 0.24254217743873596,
+      "learning_rate": 9.365596713521077e-05,
+      "loss": 0.0976,
+      "step": 1220
+    },
+    {
+      "epoch": 1.720531654424624,
+      "grad_norm": 0.34802696108818054,
+      "learning_rate": 9.203239002817377e-05,
+      "loss": 0.0916,
+      "step": 1230
+    },
+    {
+      "epoch": 1.734522560335782,
+      "grad_norm": 0.2737961411476135,
+      "learning_rate": 9.041092399585368e-05,
+      "loss": 0.0983,
+      "step": 1240
+    },
+    {
+      "epoch": 1.7485134662469395,
+      "grad_norm": 0.2548612654209137,
+      "learning_rate": 8.879199865716479e-05,
+      "loss": 0.0903,
+      "step": 1250
+    },
+    {
+      "epoch": 1.762504372158097,
+      "grad_norm": 0.276997834444046,
+      "learning_rate": 8.717604295784657e-05,
+      "loss": 0.099,
+      "step": 1260
+    },
+    {
+      "epoch": 1.776495278069255,
+      "grad_norm": 0.27574673295021057,
+      "learning_rate": 8.556348505681147e-05,
+      "loss": 0.0935,
+      "step": 1270
+    },
+    {
+      "epoch": 1.7904861839804127,
+      "grad_norm": 0.27950429916381836,
+      "learning_rate": 8.395475221270124e-05,
+      "loss": 0.1004,
+      "step": 1280
+    },
+    {
+      "epoch": 1.8044770898915705,
+      "grad_norm": 0.3705288767814636,
+      "learning_rate": 8.235027067068175e-05,
+      "loss": 0.1005,
+      "step": 1290
+    },
+    {
+      "epoch": 1.8184679958027283,
+      "grad_norm": 0.23482586443424225,
+      "learning_rate": 8.075046554950612e-05,
+      "loss": 0.0952,
+      "step": 1300
+    },
+    {
+      "epoch": 1.832458901713886,
+      "grad_norm": 0.2208295315504074,
+      "learning_rate": 7.91557607288767e-05,
+      "loss": 0.095,
+      "step": 1310
+    },
+    {
+      "epoch": 1.8464498076250437,
+      "grad_norm": 0.3105419874191284,
+      "learning_rate": 7.756657873713513e-05,
+      "loss": 0.0817,
+      "step": 1320
+    },
+    {
+      "epoch": 1.8604407135362013,
+      "grad_norm": 0.2837795615196228,
+      "learning_rate": 7.598334063931032e-05,
+      "loss": 0.0961,
+      "step": 1330
+    },
+    {
+      "epoch": 1.8744316194473591,
+      "grad_norm": 0.25267350673675537,
+      "learning_rate": 7.440646592555442e-05,
+      "loss": 0.0954,
+      "step": 1340
+    },
+    {
+      "epoch": 1.888422525358517,
+      "grad_norm": 0.23757292330265045,
+      "learning_rate": 7.28363723999959e-05,
+      "loss": 0.0766,
+      "step": 1350
+    },
+    {
+      "epoch": 1.9024134312696748,
+      "grad_norm": 0.27169138193130493,
+      "learning_rate": 7.127347607003929e-05,
+      "loss": 0.0879,
+      "step": 1360
+    },
+    {
+      "epoch": 1.9164043371808326,
+      "grad_norm": 0.2501109540462494,
+      "learning_rate": 6.971819103614118e-05,
+      "loss": 0.0881,
+      "step": 1370
+    },
+    {
+      "epoch": 1.9303952430919902,
+      "grad_norm": 0.2925432324409485,
+      "learning_rate": 6.81709293820914e-05,
+      "loss": 0.0927,
+      "step": 1380
+    },
+    {
+      "epoch": 1.944386149003148,
+      "grad_norm": 0.29960620403289795,
+      "learning_rate": 6.663210106582838e-05,
+      "loss": 0.1028,
+      "step": 1390
+    },
+    {
+      "epoch": 1.9583770549143056,
+      "grad_norm": 0.28269627690315247,
+      "learning_rate": 6.510211381081794e-05,
+      "loss": 0.0961,
+      "step": 1400
+    },
+    {
+      "epoch": 1.9583770549143056,
+      "eval_loss": 0.10136425495147705,
+      "eval_runtime": 22.0658,
+      "eval_samples_per_second": 14.411,
+      "eval_steps_per_second": 3.626,
+      "step": 1400
+    },
+    {
+      "epoch": 1.9723679608254634,
+      "grad_norm": 0.261138379573822,
+      "learning_rate": 6.358137299802416e-05,
+      "loss": 0.0869,
+      "step": 1410
+    },
+    {
+      "epoch": 1.9863588667366212,
+      "grad_norm": 0.2551378309726715,
+      "learning_rate": 6.207028155850062e-05,
+      "loss": 0.1001,
+      "step": 1420
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.26610472798347473,
+      "learning_rate": 6.0569239866631124e-05,
+      "loss": 0.1051,
+      "step": 1430
+    },
+    {
+      "epoch": 2.013990905911158,
+      "grad_norm": 0.3179965317249298,
+      "learning_rate": 5.90786456340477e-05,
+      "loss": 0.0866,
+      "step": 1440
+    },
+    {
+      "epoch": 2.0279818118223156,
+      "grad_norm": 0.21667960286140442,
+      "learning_rate": 5.759889380425393e-05,
+      "loss": 0.0839,
+      "step": 1450
+    },
+    {
+      "epoch": 2.0419727177334734,
+      "grad_norm": 0.26058515906333923,
+      "learning_rate": 5.613037644798196e-05,
+      "loss": 0.0659,
+      "step": 1460
+    },
+    {
+      "epoch": 2.055963623644631,
+      "grad_norm": 0.2897161841392517,
+      "learning_rate": 5.467348265931063e-05,
+      "loss": 0.0785,
+      "step": 1470
+    },
+    {
+      "epoch": 2.0699545295557886,
+      "grad_norm": 0.24778111279010773,
+      "learning_rate": 5.3228598452572084e-05,
+      "loss": 0.0743,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0839454354669464,
+      "grad_norm": 0.2585526704788208,
+      "learning_rate": 5.179610666007475e-05,
+      "loss": 0.0844,
+      "step": 1490
+    },
+    {
+      "epoch": 2.0979363413781043,
+      "grad_norm": 0.27612507343292236,
+      "learning_rate": 5.037638683066903e-05,
+      "loss": 0.0978,
+      "step": 1500
+    },
+    {
+      "epoch": 2.111927247289262,
+      "grad_norm": 0.2550652325153351,
+      "learning_rate": 4.896981512918325e-05,
+      "loss": 0.0788,
+      "step": 1510
+    },
+    {
+      "epoch": 2.12591815320042,
+      "grad_norm": 0.33637735247612,
+      "learning_rate": 4.757676423675601e-05,
+      "loss": 0.0833,
+      "step": 1520
+    },
+    {
+      "epoch": 2.1399090591115772,
+      "grad_norm": 0.28613343834877014,
+      "learning_rate": 4.619760325209184e-05,
+      "loss": 0.0786,
+      "step": 1530
+    },
+    {
+      "epoch": 2.153899965022735,
+      "grad_norm": 0.22645096480846405,
+      "learning_rate": 4.483269759366569e-05,
+      "loss": 0.0726,
+      "step": 1540
+    },
+    {
+      "epoch": 2.167890870933893,
+      "grad_norm": 0.2636771500110626,
+      "learning_rate": 4.348240890290289e-05,
+      "loss": 0.0794,
+      "step": 1550
+    },
+    {
+      "epoch": 2.1818817768450507,
+      "grad_norm": 0.312181293964386,
+      "learning_rate": 4.214709494835962e-05,
+      "loss": 0.077,
+      "step": 1560
+    },
+    {
+      "epoch": 2.1958726827562085,
+      "grad_norm": 0.32286760210990906,
+      "learning_rate": 4.0827109530929595e-05,
+      "loss": 0.0798,
+      "step": 1570
+    },
+    {
+      "epoch": 2.2098635886673663,
+      "grad_norm": 0.3038155138492584,
+      "learning_rate": 3.952280239010192e-05,
+      "loss": 0.0802,
+      "step": 1580
+    },
+    {
+      "epoch": 2.223854494578524,
+      "grad_norm": 0.28648459911346436,
+      "learning_rate": 3.8234519111295365e-05,
+      "loss": 0.0833,
+      "step": 1590
+    },
+    {
+      "epoch": 2.2378454004896815,
+      "grad_norm": 0.32507434487342834,
+      "learning_rate": 3.6962601034292766e-05,
+      "loss": 0.076,
+      "step": 1600
+    },
+    {
+      "epoch": 2.2378454004896815,
+      "eval_loss": 0.09979136288166046,
+      "eval_runtime": 22.3104,
+      "eval_samples_per_second": 14.253,
+      "eval_steps_per_second": 3.586,
+      "step": 1600
+    },
+    {
+      "epoch": 2.2518363064008393,
+      "grad_norm": 0.25779277086257935,
+      "learning_rate": 3.570738516280092e-05,
+      "loss": 0.0798,
+      "step": 1610
+    },
+    {
+      "epoch": 2.265827212311997,
+      "grad_norm": 0.27955761551856995,
+      "learning_rate": 3.446920407515887e-05,
+      "loss": 0.0758,
+      "step": 1620
+    },
+    {
+      "epoch": 2.279818118223155,
+      "grad_norm": 0.3353288471698761,
+      "learning_rate": 3.3248385836218977e-05,
+      "loss": 0.0803,
+      "step": 1630
+    },
+    {
+      "epoch": 2.2938090241343128,
+      "grad_norm": 0.2533150613307953,
+      "learning_rate": 3.2045253910423864e-05,
+      "loss": 0.0754,
+      "step": 1640
+    },
+    {
+      "epoch": 2.3077999300454706,
+      "grad_norm": 0.30213606357574463,
+      "learning_rate": 3.086012707610209e-05,
+      "loss": 0.0726,
+      "step": 1650
+    },
+    {
+      "epoch": 2.3217908359566284,
+      "grad_norm": 0.3189963400363922,
+      "learning_rate": 2.9693319341005633e-05,
+      "loss": 0.0798,
+      "step": 1660
+    },
+    {
+      "epoch": 2.3357817418677858,
+      "grad_norm": 0.33093076944351196,
+      "learning_rate": 2.8545139859111446e-05,
+      "loss": 0.0719,
+      "step": 1670
+    },
+    {
+      "epoch": 2.3497726477789436,
+      "grad_norm": 0.25722455978393555,
+      "learning_rate": 2.7415892848708725e-05,
+      "loss": 0.069,
+      "step": 1680
+    },
+    {
+      "epoch": 2.3637635536901014,
+      "grad_norm": 0.30695757269859314,
+      "learning_rate": 2.630587751179422e-05,
+      "loss": 0.0822,
+      "step": 1690
+    },
+    {
+      "epoch": 2.377754459601259,
+      "grad_norm": 0.30474889278411865,
+      "learning_rate": 2.5215387954796555e-05,
+      "loss": 0.0744,
+      "step": 1700
+    },
+    {
+      "epoch": 2.391745365512417,
+      "grad_norm": 0.2297676056623459,
+      "learning_rate": 2.414471311065053e-05,
+      "loss": 0.0695,
+      "step": 1710
+    },
+    {
+      "epoch": 2.405736271423575,
+      "grad_norm": 0.4082636535167694,
+      "learning_rate": 2.3094136662242405e-05,
+      "loss": 0.0792,
+      "step": 1720
+    },
+    {
+      "epoch": 2.4197271773347326,
+      "grad_norm": 0.3242763876914978,
+      "learning_rate": 2.2063936967246136e-05,
+      "loss": 0.0823,
+      "step": 1730
+    },
+    {
+      "epoch": 2.43371808324589,
+      "grad_norm": 0.28933578729629517,
+      "learning_rate": 2.1054386984370434e-05,
+      "loss": 0.0768,
+      "step": 1740
+    },
+    {
+      "epoch": 2.447708989157048,
+      "grad_norm": 0.3418210744857788,
+      "learning_rate": 2.006575420103648e-05,
+      "loss": 0.0729,
+      "step": 1750
+    },
+    {
+      "epoch": 2.4616998950682056,
+      "grad_norm": 0.33131641149520874,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.0851,
+      "step": 1760
+    },
+    {
+      "epoch": 2.4756908009793634,
+      "grad_norm": 0.2488657683134079,
+      "learning_rate": 1.815228240247323e-05,
+      "loss": 0.0703,
+      "step": 1770
+    },
+    {
+      "epoch": 2.4896817068905213,
+      "grad_norm": 0.2939601540565491,
+      "learning_rate": 1.7227950375154912e-05,
+      "loss": 0.068,
+      "step": 1780
+    },
+    {
+      "epoch": 2.5036726128016786,
+      "grad_norm": 0.2726001441478729,
+      "learning_rate": 1.6325549388870365e-05,
+      "loss": 0.0673,
+      "step": 1790
+    },
+    {
+      "epoch": 2.517663518712837,
+      "grad_norm": 0.2877500653266907,
+      "learning_rate": 1.5445318541154942e-05,
+      "loss": 0.0722,
+      "step": 1800
+    },
+    {
+      "epoch": 2.517663518712837,
+      "eval_loss": 0.09834001958370209,
+      "eval_runtime": 21.9639,
+      "eval_samples_per_second": 14.478,
+      "eval_steps_per_second": 3.642,
+      "step": 1800
+    },
+    {
+      "epoch": 2.5316544246239943,
+      "grad_norm": 0.29231616854667664,
+      "learning_rate": 1.458749105540863e-05,
+      "loss": 0.0809,
+      "step": 1810
+    },
+    {
+      "epoch": 2.545645330535152,
+      "grad_norm": 0.3287057876586914,
+      "learning_rate": 1.3752294219102025e-05,
+      "loss": 0.0666,
+      "step": 1820
+    },
+    {
+      "epoch": 2.55963623644631,
+      "grad_norm": 0.32760322093963623,
+      "learning_rate": 1.2939949323554668e-05,
+      "loss": 0.0773,
+      "step": 1830
+    },
+    {
+      "epoch": 2.5736271423574677,
+      "grad_norm": 0.3478274643421173,
+      "learning_rate": 1.2150671605302566e-05,
+      "loss": 0.0777,
+      "step": 1840
+    },
+    {
+      "epoch": 2.5876180482686255,
+      "grad_norm": 0.29048407077789307,
+      "learning_rate": 1.1384670189069635e-05,
+      "loss": 0.078,
+      "step": 1850
+    },
+    {
+      "epoch": 2.601608954179783,
+      "grad_norm": 0.36669981479644775,
+      "learning_rate": 1.0642148032358645e-05,
+      "loss": 0.084,
+      "step": 1860
+    },
+    {
+      "epoch": 2.615599860090941,
+      "grad_norm": 0.3382461667060852,
+      "learning_rate": 9.923301871676227e-06,
+      "loss": 0.0758,
+      "step": 1870
+    },
+    {
+      "epoch": 2.6295907660020985,
+      "grad_norm": 0.3543117344379425,
+      "learning_rate": 9.228322170406056e-06,
+      "loss": 0.0665,
+      "step": 1880
+    },
+    {
+      "epoch": 2.6435816719132563,
+      "grad_norm": 0.4150249660015106,
+      "learning_rate": 8.55739306834431e-06,
+      "loss": 0.0782,
+      "step": 1890
+    },
+    {
+      "epoch": 2.657572577824414,
+      "grad_norm": 0.2844158113002777,
+      "learning_rate": 7.910692332910452e-06,
+      "loss": 0.0723,
+      "step": 1900
+    },
+    {
+      "epoch": 2.671563483735572,
+      "grad_norm": 0.39910125732421875,
+      "learning_rate": 7.288391312046583e-06,
+      "loss": 0.0731,
+      "step": 1910
+    },
+    {
+      "epoch": 2.6855543896467298,
+      "grad_norm": 0.2358187735080719,
+      "learning_rate": 6.690654888817538e-06,
+      "loss": 0.0697,
+      "step": 1920
+    },
+    {
+      "epoch": 2.699545295557887,
+      "grad_norm": 0.3523266017436981,
+      "learning_rate": 6.117641437723931e-06,
+      "loss": 0.0781,
+      "step": 1930
+    },
+    {
+      "epoch": 2.7135362014690454,
+      "grad_norm": 0.3880462348461151,
+      "learning_rate": 5.569502782739766e-06,
+      "loss": 0.0718,
+      "step": 1940
+    },
+    {
+      "epoch": 2.7275271073802028,
+      "grad_norm": 0.3129269778728485,
+      "learning_rate": 5.046384157085537e-06,
+      "loss": 0.0679,
+      "step": 1950
+    },
+    {
+      "epoch": 2.7415180132913606,
+      "grad_norm": 0.33428144454956055,
+      "learning_rate": 4.548424164747667e-06,
+      "loss": 0.069,
+      "step": 1960
+    },
+    {
+      "epoch": 2.7555089192025184,
+      "grad_norm": 0.30948323011398315,
+      "learning_rate": 4.075754743754478e-06,
+      "loss": 0.0625,
+      "step": 1970
+    },
+    {
+      "epoch": 2.769499825113676,
+      "grad_norm": 0.23879970610141754,
+      "learning_rate": 3.6285011312181873e-06,
+      "loss": 0.0801,
+      "step": 1980
+    },
+    {
+      "epoch": 2.783490731024834,
+      "grad_norm": 0.3375139832496643,
+      "learning_rate": 3.2067818301524497e-06,
+      "loss": 0.082,
+      "step": 1990
+    },
+    {
+      "epoch": 2.7974816369359914,
+      "grad_norm": 0.35570380091667175,
+      "learning_rate": 2.810708578074239e-06,
+      "loss": 0.0717,
+      "step": 2000
+    },
+    {
+      "epoch": 2.7974816369359914,
+      "eval_loss": 0.09761384129524231,
+      "eval_runtime": 21.573,
+      "eval_samples_per_second": 14.741,
+      "eval_steps_per_second": 3.708,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2145,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.552876470153216e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoints/checkpoint-2000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbc9bc83f818f5138562feaa20da67402920a7f2ed95028654b04bd38ebe547e
+size 6417

checkpoints/checkpoint-2000/video_preprocessor_config.json

@@ -0,0 +1,41 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": true,
+  "device": null,
+  "do_center_crop": null,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "do_sample_frames": true,
+  "fps": 2,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "input_data_format": null,
+  "max_frames": 768,
+  "merge_size": 2,
+  "min_frames": 4,
+  "num_frames": null,
+  "pad_size": null,
+  "patch_size": 16,
+  "processor_class": "Qwen3VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "return_metadata": false,
+  "size": {
+    "longest_edge": 25165824,
+    "shortest_edge": 4096
+  },
+  "temporal_patch_size": 2,
+  "video_metadata": null,
+  "video_processor_type": "Qwen3VLVideoProcessor"
+}

checkpoints/checkpoint-2145/README.md

@@ -0,0 +1,210 @@
+---
+base_model: unsloth/qwen3-vl-2b-instruct-unsloth-bnb-4bit
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:unsloth/qwen3-vl-2b-instruct-unsloth-bnb-4bit
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

checkpoints/checkpoint-2145/adapter_config.json

@@ -0,0 +1,50 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Qwen3VLForConditionalGeneration",
+    "parent_library": "transformers.models.qwen3_vl.modeling_qwen3_vl",
+    "unsloth_fixed": true
+  },
+  "base_model_name_or_path": "unsloth/qwen3-vl-2b-instruct-unsloth-bnb-4bit",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "o_proj",
+    "down_proj",
+    "q_proj",
+    "gate_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

checkpoints/checkpoint-2145/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd296c2718a6c1b83cc3b19cc8ccf38f2f3b46f5336d1866a4a0f99f3173afae
+size 69788264

checkpoints/checkpoint-2145/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

checkpoints/checkpoint-2145/chat_template.jinja

@@ -0,0 +1,120 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' }}
+        {%- if messages[0].content is string %}
+            {{- messages[0].content }}
+        {%- else %}
+            {%- for content in messages[0].content %}
+                {%- if 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- for message in messages %}
+    {%- if message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content_item in message.content %}
+                {%- if 'text' in content_item %}
+                    {{- content_item.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and message.content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {%- if message.content is string %}
+            {{- message.content }}
+        {%- else %}
+            {%- for content in message.content %}
+                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}
+                    <|vision_start|><|image_pad|><|vision_end|>
+                {%- elif content.type == 'video' or 'video' in content %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}
+                    <|vision_start|><|video_pad|><|vision_end|>
+                {%- elif 'text' in content %}
+                    {{- content.text }}
+                {%- endif %}
+            {%- endfor %}
+        {%- endif %}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

checkpoints/checkpoint-2145/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc1aed833c3fcca33e49277a7f61582e0d954ba0615045857f23ed9333f26a85
+size 35869445

checkpoints/checkpoint-2145/preprocessor_config.json

@@ -0,0 +1,39 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": true,
+  "device": null,
+  "disable_grouping": null,
+  "do_center_crop": null,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_pad": null,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "Qwen2VLImageProcessorFast",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "input_data_format": null,
+  "max_pixels": null,
+  "merge_size": 2,
+  "min_pixels": null,
+  "pad_size": null,
+  "patch_size": 16,
+  "processor_class": "Qwen3VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "return_tensors": null,
+  "size": {
+    "longest_edge": 16777216,
+    "shortest_edge": 65536
+  },
+  "temporal_patch_size": 2
+}

checkpoints/checkpoint-2145/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89f2e7b9901fa4b72af37f5845fbfc651b4a741e105cc6dfc15f2793ce6c6f67
+size 14645

checkpoints/checkpoint-2145/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d653fe7544acebc73dbd211d67a88bc7f377ae0fc640a5549094b1d4a317837c
+size 1465

checkpoints/checkpoint-2145/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|vision_pad|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoints/checkpoint-2145/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af560c5e5807192776a9f72a0ea33c9331816d9c59af61e02bdf766f8a2a97fb
+size 11422922

checkpoints/checkpoint-2145/tokenizer_config.json

@@ -0,0 +1,241 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 262144,
+  "pad_token": "<|vision_pad|>",
+  "padding_side": "right",
+  "processor_class": "Qwen3VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

checkpoints/checkpoint-2145/trainer_state.json

@@ -0,0 +1,1612 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 200,
+  "global_step": 2145,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.013990905911157748,
+      "grad_norm": 11.852328300476074,
+      "learning_rate": 8.372093023255815e-06,
+      "loss": 3.4227,
+      "step": 10
+    },
+    {
+      "epoch": 0.027981811822315496,
+      "grad_norm": 5.509212017059326,
+      "learning_rate": 1.7674418604651163e-05,
+      "loss": 3.1068,
+      "step": 20
+    },
+    {
+      "epoch": 0.04197271773347324,
+      "grad_norm": 3.556685209274292,
+      "learning_rate": 2.6976744186046517e-05,
+      "loss": 2.3908,
+      "step": 30
+    },
+    {
+      "epoch": 0.05596362364463099,
+      "grad_norm": 4.118677616119385,
+      "learning_rate": 3.627906976744187e-05,
+      "loss": 1.5286,
+      "step": 40
+    },
+    {
+      "epoch": 0.06995452955578874,
+      "grad_norm": 2.5455708503723145,
+      "learning_rate": 4.5581395348837214e-05,
+      "loss": 0.7724,
+      "step": 50
+    },
+    {
+      "epoch": 0.08394543546694648,
+      "grad_norm": 1.4716579914093018,
+      "learning_rate": 5.488372093023256e-05,
+      "loss": 0.3533,
+      "step": 60
+    },
+    {
+      "epoch": 0.09793634137810424,
+      "grad_norm": 1.154810905456543,
+      "learning_rate": 6.418604651162791e-05,
+      "loss": 0.2292,
+      "step": 70
+    },
+    {
+      "epoch": 0.11192724728926198,
+      "grad_norm": 0.8629467487335205,
+      "learning_rate": 7.348837209302326e-05,
+      "loss": 0.1772,
+      "step": 80
+    },
+    {
+      "epoch": 0.1259181532004197,
+      "grad_norm": 0.7331057190895081,
+      "learning_rate": 8.27906976744186e-05,
+      "loss": 0.1643,
+      "step": 90
+    },
+    {
+      "epoch": 0.13990905911157747,
+      "grad_norm": 0.5298661589622498,
+      "learning_rate": 9.209302325581396e-05,
+      "loss": 0.1659,
+      "step": 100
+    },
+    {
+      "epoch": 0.15389996502273523,
+      "grad_norm": 0.6423642635345459,
+      "learning_rate": 0.00010139534883720931,
+      "loss": 0.1569,
+      "step": 110
+    },
+    {
+      "epoch": 0.16789087093389296,
+      "grad_norm": 0.7554107904434204,
+      "learning_rate": 0.00011069767441860466,
+      "loss": 0.1335,
+      "step": 120
+    },
+    {
+      "epoch": 0.18188177684505072,
+      "grad_norm": 0.6365725994110107,
+      "learning_rate": 0.00012,
+      "loss": 0.1346,
+      "step": 130
+    },
+    {
+      "epoch": 0.19587268275620848,
+      "grad_norm": 0.577471137046814,
+      "learning_rate": 0.00012930232558139534,
+      "loss": 0.1559,
+      "step": 140
+    },
+    {
+      "epoch": 0.2098635886673662,
+      "grad_norm": 0.6245256066322327,
+      "learning_rate": 0.00013860465116279072,
+      "loss": 0.1678,
+      "step": 150
+    },
+    {
+      "epoch": 0.22385449457852397,
+      "grad_norm": 0.5085740089416504,
+      "learning_rate": 0.00014790697674418605,
+      "loss": 0.162,
+      "step": 160
+    },
+    {
+      "epoch": 0.2378454004896817,
+      "grad_norm": 0.46424826979637146,
+      "learning_rate": 0.0001572093023255814,
+      "loss": 0.1328,
+      "step": 170
+    },
+    {
+      "epoch": 0.2518363064008394,
+      "grad_norm": 0.4451293647289276,
+      "learning_rate": 0.00016651162790697673,
+      "loss": 0.1528,
+      "step": 180
+    },
+    {
+      "epoch": 0.2658272123119972,
+      "grad_norm": 0.6359198689460754,
+      "learning_rate": 0.0001758139534883721,
+      "loss": 0.1334,
+      "step": 190
+    },
+    {
+      "epoch": 0.27981811822315494,
+      "grad_norm": 0.406089723110199,
+      "learning_rate": 0.00018511627906976744,
+      "loss": 0.1414,
+      "step": 200
+    },
+    {
+      "epoch": 0.27981811822315494,
+      "eval_loss": 0.1422451138496399,
+      "eval_runtime": 21.5015,
+      "eval_samples_per_second": 14.79,
+      "eval_steps_per_second": 3.721,
+      "step": 200
+    },
+    {
+      "epoch": 0.2938090241343127,
+      "grad_norm": 0.4091590642929077,
+      "learning_rate": 0.0001944186046511628,
+      "loss": 0.1526,
+      "step": 210
+    },
+    {
+      "epoch": 0.30779993004547046,
+      "grad_norm": 0.325090229511261,
+      "learning_rate": 0.000199997880304001,
+      "loss": 0.1173,
+      "step": 220
+    },
+    {
+      "epoch": 0.3217908359566282,
+      "grad_norm": 0.3782133460044861,
+      "learning_rate": 0.0001999740347560115,
+      "loss": 0.1445,
+      "step": 230
+    },
+    {
+      "epoch": 0.3357817418677859,
+      "grad_norm": 0.4928087592124939,
+      "learning_rate": 0.00019992370037914218,
+      "loss": 0.1363,
+      "step": 240
+    },
+    {
+      "epoch": 0.3497726477789437,
+      "grad_norm": 0.31038427352905273,
+      "learning_rate": 0.00019984689050984278,
+      "loss": 0.1296,
+      "step": 250
+    },
+    {
+      "epoch": 0.36376355369010144,
+      "grad_norm": 0.4541214108467102,
+      "learning_rate": 0.00019974362549943247,
+      "loss": 0.1207,
+      "step": 260
+    },
+    {
+      "epoch": 0.3777544596012592,
+      "grad_norm": 0.46741530299186707,
+      "learning_rate": 0.00019961393270870738,
+      "loss": 0.1386,
+      "step": 270
+    },
+    {
+      "epoch": 0.39174536551241695,
+      "grad_norm": 0.3349354863166809,
+      "learning_rate": 0.00019945784650069143,
+      "loss": 0.1533,
+      "step": 280
+    },
+    {
+      "epoch": 0.40573627142357466,
+      "grad_norm": 0.40393367409706116,
+      "learning_rate": 0.00019927540823153135,
+      "loss": 0.135,
+      "step": 290
+    },
+    {
+      "epoch": 0.4197271773347324,
+      "grad_norm": 0.34774526953697205,
+      "learning_rate": 0.00019906666623953925,
+      "loss": 0.1213,
+      "step": 300
+    },
+    {
+      "epoch": 0.4337180832458902,
+      "grad_norm": 0.35436466336250305,
+      "learning_rate": 0.00019883167583238494,
+      "loss": 0.1244,
+      "step": 310
+    },
+    {
+      "epoch": 0.44770898915704793,
+      "grad_norm": 0.3164362609386444,
+      "learning_rate": 0.00019857049927244185,
+      "loss": 0.1243,
+      "step": 320
+    },
+    {
+      "epoch": 0.4616998950682057,
+      "grad_norm": 0.3490104675292969,
+      "learning_rate": 0.0001982832057602901,
+      "loss": 0.1427,
+      "step": 330
+    },
+    {
+      "epoch": 0.4756908009793634,
+      "grad_norm": 0.4742378294467926,
+      "learning_rate": 0.00019796987141638134,
+      "loss": 0.1252,
+      "step": 340
+    },
+    {
+      "epoch": 0.48968170689052115,
+      "grad_norm": 0.33545687794685364,
+      "learning_rate": 0.00019763057926087022,
+      "loss": 0.1355,
+      "step": 350
+    },
+    {
+      "epoch": 0.5036726128016789,
+      "grad_norm": 0.4260849058628082,
+      "learning_rate": 0.00019726541919161733,
+      "loss": 0.1125,
+      "step": 360
+    },
+    {
+      "epoch": 0.5176635187128367,
+      "grad_norm": 0.4119410514831543,
+      "learning_rate": 0.00019687448796037045,
+      "loss": 0.122,
+      "step": 370
+    },
+    {
+      "epoch": 0.5316544246239944,
+      "grad_norm": 0.28887462615966797,
+      "learning_rate": 0.0001964578891471292,
+      "loss": 0.129,
+      "step": 380
+    },
+    {
+      "epoch": 0.5456453305351522,
+      "grad_norm": 0.3469299077987671,
+      "learning_rate": 0.00019601573313270102,
+      "loss": 0.1211,
+      "step": 390
+    },
+    {
+      "epoch": 0.5596362364463099,
+      "grad_norm": 0.2652361989021301,
+      "learning_rate": 0.00019554813706945486,
+      "loss": 0.1217,
+      "step": 400
+    },
+    {
+      "epoch": 0.5596362364463099,
+      "eval_loss": 0.12272346764802933,
+      "eval_runtime": 21.4783,
+      "eval_samples_per_second": 14.806,
+      "eval_steps_per_second": 3.725,
+      "step": 400
+    },
+    {
+      "epoch": 0.5736271423574677,
+      "grad_norm": 0.4036720395088196,
+      "learning_rate": 0.0001950552248502808,
+      "loss": 0.1341,
+      "step": 410
+    },
+    {
+      "epoch": 0.5876180482686254,
+      "grad_norm": 0.4273042380809784,
+      "learning_rate": 0.00019453712707576392,
+      "loss": 0.1093,
+      "step": 420
+    },
+    {
+      "epoch": 0.6016089541797831,
+      "grad_norm": 0.3455790877342224,
+      "learning_rate": 0.00019399398101958056,
+      "loss": 0.1133,
+      "step": 430
+    },
+    {
+      "epoch": 0.6155998600909409,
+      "grad_norm": 0.2522144019603729,
+      "learning_rate": 0.0001934259305921268,
+      "loss": 0.1221,
+      "step": 440
+    },
+    {
+      "epoch": 0.6295907660020986,
+      "grad_norm": 0.21513080596923828,
+      "learning_rate": 0.00019283312630238836,
+      "loss": 0.1112,
+      "step": 450
+    },
+    {
+      "epoch": 0.6435816719132564,
+      "grad_norm": 0.2985060214996338,
+      "learning_rate": 0.00019221572521806212,
+      "loss": 0.1072,
+      "step": 460
+    },
+    {
+      "epoch": 0.6575725778244141,
+      "grad_norm": 0.23942740261554718,
+      "learning_rate": 0.00019157389092394004,
+      "loss": 0.113,
+      "step": 470
+    },
+    {
+      "epoch": 0.6715634837355718,
+      "grad_norm": 0.35044464468955994,
+      "learning_rate": 0.00019090779347856606,
+      "loss": 0.1186,
+      "step": 480
+    },
+    {
+      "epoch": 0.6855543896467297,
+      "grad_norm": 0.25626683235168457,
+      "learning_rate": 0.00019021760936917804,
+      "loss": 0.1156,
+      "step": 490
+    },
+    {
+      "epoch": 0.6995452955578874,
+      "grad_norm": 0.3051966428756714,
+      "learning_rate": 0.00018950352146494606,
+      "loss": 0.1166,
+      "step": 500
+    },
+    {
+      "epoch": 0.7135362014690452,
+      "grad_norm": 0.3047504127025604,
+      "learning_rate": 0.00018876571896852005,
+      "loss": 0.1155,
+      "step": 510
+    },
+    {
+      "epoch": 0.7275271073802029,
+      "grad_norm": 0.2766532301902771,
+      "learning_rate": 0.00018800439736589924,
+      "loss": 0.1144,
+      "step": 520
+    },
+    {
+      "epoch": 0.7415180132913606,
+      "grad_norm": 0.321286141872406,
+      "learning_rate": 0.00018721975837463655,
+      "loss": 0.1032,
+      "step": 530
+    },
+    {
+      "epoch": 0.7555089192025184,
+      "grad_norm": 0.27030113339424133,
+      "learning_rate": 0.00018641200989039236,
+      "loss": 0.1201,
+      "step": 540
+    },
+    {
+      "epoch": 0.7694998251136761,
+      "grad_norm": 0.3184170424938202,
+      "learning_rate": 0.00018558136593185086,
+      "loss": 0.1234,
+      "step": 550
+    },
+    {
+      "epoch": 0.7834907310248339,
+      "grad_norm": 0.2622183859348297,
+      "learning_rate": 0.00018472804658401453,
+      "loss": 0.1015,
+      "step": 560
+    },
+    {
+      "epoch": 0.7974816369359916,
+      "grad_norm": 0.3095245659351349,
+      "learning_rate": 0.00018385227793989075,
+      "loss": 0.122,
+      "step": 570
+    },
+    {
+      "epoch": 0.8114725428471493,
+      "grad_norm": 0.4251258969306946,
+      "learning_rate": 0.00018295429204058712,
+      "loss": 0.1167,
+      "step": 580
+    },
+    {
+      "epoch": 0.8254634487583071,
+      "grad_norm": 0.31186479330062866,
+      "learning_rate": 0.00018203432681383034,
+      "loss": 0.1114,
+      "step": 590
+    },
+    {
+      "epoch": 0.8394543546694648,
+      "grad_norm": 0.2990236282348633,
+      "learning_rate": 0.00018109262601092576,
+      "loss": 0.1103,
+      "step": 600
+    },
+    {
+      "epoch": 0.8394543546694648,
+      "eval_loss": 0.11650671809911728,
+      "eval_runtime": 21.574,
+      "eval_samples_per_second": 14.74,
+      "eval_steps_per_second": 3.708,
+      "step": 600
+    },
+    {
+      "epoch": 0.8534452605806226,
+      "grad_norm": 0.3199589252471924,
+      "learning_rate": 0.00018012943914217356,
+      "loss": 0.1098,
+      "step": 610
+    },
+    {
+      "epoch": 0.8674361664917803,
+      "grad_norm": 0.2671171724796295,
+      "learning_rate": 0.00017914502141075928,
+      "loss": 0.0959,
+      "step": 620
+    },
+    {
+      "epoch": 0.881427072402938,
+      "grad_norm": 0.371975302696228,
+      "learning_rate": 0.00017813963364513607,
+      "loss": 0.1089,
+      "step": 630
+    },
+    {
+      "epoch": 0.8954179783140959,
+      "grad_norm": 0.28194791078567505,
+      "learning_rate": 0.00017711354222991606,
+      "loss": 0.1009,
+      "step": 640
+    },
+    {
+      "epoch": 0.9094088842252536,
+      "grad_norm": 0.24068565666675568,
+      "learning_rate": 0.0001760670190352903,
+      "loss": 0.1195,
+      "step": 650
+    },
+    {
+      "epoch": 0.9233997901364114,
+      "grad_norm": 0.35747045278549194,
+      "learning_rate": 0.00017500034134499423,
+      "loss": 0.1237,
+      "step": 660
+    },
+    {
+      "epoch": 0.9373906960475691,
+      "grad_norm": 0.30066290497779846,
+      "learning_rate": 0.00017391379178284002,
+      "loss": 0.1231,
+      "step": 670
+    },
+    {
+      "epoch": 0.9513816019587268,
+      "grad_norm": 0.3015068769454956,
+      "learning_rate": 0.0001728076582378329,
+      "loss": 0.1,
+      "step": 680
+    },
+    {
+      "epoch": 0.9653725078698846,
+      "grad_norm": 0.3291875422000885,
+      "learning_rate": 0.0001716822337878932,
+      "loss": 0.1256,
+      "step": 690
+    },
+    {
+      "epoch": 0.9793634137810423,
+      "grad_norm": 0.2612892985343933,
+      "learning_rate": 0.0001705378166222032,
+      "loss": 0.1114,
+      "step": 700
+    },
+    {
+      "epoch": 0.9933543196922001,
+      "grad_norm": 0.24723578989505768,
+      "learning_rate": 0.00016937470996219973,
+      "loss": 0.1021,
+      "step": 710
+    },
+    {
+      "epoch": 1.006995452955579,
+      "grad_norm": 0.21942616999149323,
+      "learning_rate": 0.00016819322198123365,
+      "loss": 0.1038,
+      "step": 720
+    },
+    {
+      "epoch": 1.0209863588667367,
+      "grad_norm": 0.2657986581325531,
+      "learning_rate": 0.00016699366572291702,
+      "loss": 0.0924,
+      "step": 730
+    },
+    {
+      "epoch": 1.0349772647778943,
+      "grad_norm": 0.3046330213546753,
+      "learning_rate": 0.00016577635901818,
+      "loss": 0.095,
+      "step": 740
+    },
+    {
+      "epoch": 1.0489681706890521,
+      "grad_norm": 0.29112890362739563,
+      "learning_rate": 0.0001645416244010593,
+      "loss": 0.0939,
+      "step": 750
+    },
+    {
+      "epoch": 1.06295907660021,
+      "grad_norm": 0.2303277999162674,
+      "learning_rate": 0.00016328978902324046,
+      "loss": 0.0936,
+      "step": 760
+    },
+    {
+      "epoch": 1.0769499825113675,
+      "grad_norm": 0.21349187195301056,
+      "learning_rate": 0.00016202118456737683,
+      "loss": 0.1021,
+      "step": 770
+    },
+    {
+      "epoch": 1.0909408884225253,
+      "grad_norm": 0.2621854245662689,
+      "learning_rate": 0.0001607361471592077,
+      "loss": 0.0998,
+      "step": 780
+    },
+    {
+      "epoch": 1.1049317943336832,
+      "grad_norm": 0.21957777440547943,
+      "learning_rate": 0.0001594350172784996,
+      "loss": 0.1091,
+      "step": 790
+    },
+    {
+      "epoch": 1.1189227002448408,
+      "grad_norm": 0.27233201265335083,
+      "learning_rate": 0.0001581181396688337,
+      "loss": 0.093,
+      "step": 800
+    },
+    {
+      "epoch": 1.1189227002448408,
+      "eval_loss": 0.1152026355266571,
+      "eval_runtime": 21.6054,
+      "eval_samples_per_second": 14.719,
+      "eval_steps_per_second": 3.703,
+      "step": 800
+    },
+    {
+      "epoch": 1.1329136061559986,
+      "grad_norm": 0.3158034086227417,
+      "learning_rate": 0.00015678586324626337,
+      "loss": 0.0907,
+      "step": 810
+    },
+    {
+      "epoch": 1.1469045120671564,
+      "grad_norm": 0.30520495772361755,
+      "learning_rate": 0.00015543854100686676,
+      "loss": 0.1057,
+      "step": 820
+    },
+    {
+      "epoch": 1.1608954179783142,
+      "grad_norm": 0.21135452389717102,
+      "learning_rate": 0.0001540765299332177,
+      "loss": 0.0929,
+      "step": 830
+    },
+    {
+      "epoch": 1.1748863238894718,
+      "grad_norm": 0.32276666164398193,
+      "learning_rate": 0.00015270019089980077,
+      "loss": 0.1085,
+      "step": 840
+    },
+    {
+      "epoch": 1.1888772298006296,
+      "grad_norm": 0.27095136046409607,
+      "learning_rate": 0.00015130988857739498,
+      "loss": 0.0984,
+      "step": 850
+    },
+    {
+      "epoch": 1.2028681357117874,
+      "grad_norm": 0.4224928915500641,
+      "learning_rate": 0.0001499059913364517,
+      "loss": 0.1117,
+      "step": 860
+    },
+    {
+      "epoch": 1.216859041622945,
+      "grad_norm": 0.24559757113456726,
+      "learning_rate": 0.00014848887114949234,
+      "loss": 0.101,
+      "step": 870
+    },
+    {
+      "epoch": 1.2308499475341028,
+      "grad_norm": 0.2625027894973755,
+      "learning_rate": 0.00014705890349255149,
+      "loss": 0.1113,
+      "step": 880
+    },
+    {
+      "epoch": 1.2448408534452606,
+      "grad_norm": 0.31155043840408325,
+      "learning_rate": 0.00014561646724569198,
+      "loss": 0.0903,
+      "step": 890
+    },
+    {
+      "epoch": 1.2588317593564184,
+      "grad_norm": 0.2001536637544632,
+      "learning_rate": 0.00014416194459261785,
+      "loss": 0.123,
+      "step": 900
+    },
+    {
+      "epoch": 1.272822665267576,
+      "grad_norm": 0.23669183254241943,
+      "learning_rate": 0.00014269572091941214,
+      "loss": 0.109,
+      "step": 910
+    },
+    {
+      "epoch": 1.2868135711787339,
+      "grad_norm": 0.2512431740760803,
+      "learning_rate": 0.00014121818471242608,
+      "loss": 0.1022,
+      "step": 920
+    },
+    {
+      "epoch": 1.3008044770898914,
+      "grad_norm": 0.26644250750541687,
+      "learning_rate": 0.00013972972745534686,
+      "loss": 0.0991,
+      "step": 930
+    },
+    {
+      "epoch": 1.3147953830010493,
+      "grad_norm": 0.24515704810619354,
+      "learning_rate": 0.0001382307435254713,
+      "loss": 0.1002,
+      "step": 940
+    },
+    {
+      "epoch": 1.328786288912207,
+      "grad_norm": 0.2648073732852936,
+      "learning_rate": 0.0001367216300892127,
+      "loss": 0.1074,
+      "step": 950
+    },
+    {
+      "epoch": 1.3427771948233649,
+      "grad_norm": 0.2987382411956787,
+      "learning_rate": 0.00013520278699686897,
+      "loss": 0.1017,
+      "step": 960
+    },
+    {
+      "epoch": 1.3567681007345227,
+      "grad_norm": 0.22436612844467163,
+      "learning_rate": 0.00013367461667667923,
+      "loss": 0.0923,
+      "step": 970
+    },
+    {
+      "epoch": 1.3707590066456803,
+      "grad_norm": 0.6144154667854309,
+      "learning_rate": 0.00013213752402819773,
+      "loss": 0.1021,
+      "step": 980
+    },
+    {
+      "epoch": 1.384749912556838,
+      "grad_norm": 0.21033185720443726,
+      "learning_rate": 0.00013059191631501288,
+      "loss": 0.1136,
+      "step": 990
+    },
+    {
+      "epoch": 1.3987408184679957,
+      "grad_norm": 0.3320418894290924,
+      "learning_rate": 0.00012903820305683973,
+      "loss": 0.1065,
+      "step": 1000
+    },
+    {
+      "epoch": 1.3987408184679957,
+      "eval_loss": 0.10766011476516724,
+      "eval_runtime": 21.6633,
+      "eval_samples_per_second": 14.679,
+      "eval_steps_per_second": 3.693,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4127317243791535,
+      "grad_norm": 0.3217187523841858,
+      "learning_rate": 0.00012747679592101498,
+      "loss": 0.1025,
+      "step": 1010
+    },
+    {
+      "epoch": 1.4267226302903113,
+      "grad_norm": 0.19164179265499115,
+      "learning_rate": 0.00012590810861342266,
+      "loss": 0.0903,
+      "step": 1020
+    },
+    {
+      "epoch": 1.4407135362014691,
+      "grad_norm": 0.30840256810188293,
+      "learning_rate": 0.00012433255676888,
+      "loss": 0.0895,
+      "step": 1030
+    },
+    {
+      "epoch": 1.4547044421126267,
+      "grad_norm": 0.25482040643692017,
+      "learning_rate": 0.00012275055784101178,
+      "loss": 0.094,
+      "step": 1040
+    },
+    {
+      "epoch": 1.4686953480237845,
+      "grad_norm": 0.22336170077323914,
+      "learning_rate": 0.00012116253099164334,
+      "loss": 0.0907,
+      "step": 1050
+    },
+    {
+      "epoch": 1.4826862539349424,
+      "grad_norm": 0.3286293148994446,
+      "learning_rate": 0.00011956889697974045,
+      "loss": 0.1058,
+      "step": 1060
+    },
+    {
+      "epoch": 1.4966771598461,
+      "grad_norm": 0.28686946630477905,
+      "learning_rate": 0.00011797007804992633,
+      "loss": 0.1009,
+      "step": 1070
+    },
+    {
+      "epoch": 1.5106680657572578,
+      "grad_norm": 0.2923101782798767,
+      "learning_rate": 0.00011636649782060496,
+      "loss": 0.0993,
+      "step": 1080
+    },
+    {
+      "epoch": 1.5246589716684156,
+      "grad_norm": 0.15968987345695496,
+      "learning_rate": 0.00011475858117172034,
+      "loss": 0.0866,
+      "step": 1090
+    },
+    {
+      "epoch": 1.5386498775795734,
+      "grad_norm": 0.2587738037109375,
+      "learning_rate": 0.00011314675413218156,
+      "loss": 0.0874,
+      "step": 1100
+    },
+    {
+      "epoch": 1.552640783490731,
+      "grad_norm": 0.2577781081199646,
+      "learning_rate": 0.00011153144376698347,
+      "loss": 0.0923,
+      "step": 1110
+    },
+    {
+      "epoch": 1.5666316894018888,
+      "grad_norm": 0.27181577682495117,
+      "learning_rate": 0.00010991307806405269,
+      "loss": 0.1007,
+      "step": 1120
+    },
+    {
+      "epoch": 1.5806225953130464,
+      "grad_norm": 0.19265908002853394,
+      "learning_rate": 0.00010829208582084915,
+      "loss": 0.0997,
+      "step": 1130
+    },
+    {
+      "epoch": 1.5946135012242042,
+      "grad_norm": 0.21604153513908386,
+      "learning_rate": 0.0001066688965307534,
+      "loss": 0.0954,
+      "step": 1140
+    },
+    {
+      "epoch": 1.608604407135362,
+      "grad_norm": 0.18990111351013184,
+      "learning_rate": 0.00010504394026926902,
+      "loss": 0.0976,
+      "step": 1150
+    },
+    {
+      "epoch": 1.6225953130465198,
+      "grad_norm": 0.28930869698524475,
+      "learning_rate": 0.00010341764758007118,
+      "loss": 0.1064,
+      "step": 1160
+    },
+    {
+      "epoch": 1.6365862189576776,
+      "grad_norm": 0.3197149634361267,
+      "learning_rate": 0.0001017904493609311,
+      "loss": 0.0999,
+      "step": 1170
+    },
+    {
+      "epoch": 1.6505771248688352,
+      "grad_norm": 0.3212111294269562,
+      "learning_rate": 0.0001001627767495465,
+      "loss": 0.092,
+      "step": 1180
+    },
+    {
+      "epoch": 1.664568030779993,
+      "grad_norm": 0.1536320447921753,
+      "learning_rate": 9.853506100930875e-05,
+      "loss": 0.0979,
+      "step": 1190
+    },
+    {
+      "epoch": 1.6785589366911506,
+      "grad_norm": 0.3106030821800232,
+      "learning_rate": 9.690773341503649e-05,
+      "loss": 0.0845,
+      "step": 1200
+    },
+    {
+      "epoch": 1.6785589366911506,
+      "eval_loss": 0.10427355021238327,
+      "eval_runtime": 22.3655,
+      "eval_samples_per_second": 14.218,
+      "eval_steps_per_second": 3.577,
+      "step": 1200
+    },
+    {
+      "epoch": 1.6925498426023085,
+      "grad_norm": 0.29129135608673096,
+      "learning_rate": 9.528122513870631e-05,
+      "loss": 0.0981,
+      "step": 1210
+    },
+    {
+      "epoch": 1.7065407485134663,
+      "grad_norm": 0.24254217743873596,
+      "learning_rate": 9.365596713521077e-05,
+      "loss": 0.0976,
+      "step": 1220
+    },
+    {
+      "epoch": 1.720531654424624,
+      "grad_norm": 0.34802696108818054,
+      "learning_rate": 9.203239002817377e-05,
+      "loss": 0.0916,
+      "step": 1230
+    },
+    {
+      "epoch": 1.734522560335782,
+      "grad_norm": 0.2737961411476135,
+      "learning_rate": 9.041092399585368e-05,
+      "loss": 0.0983,
+      "step": 1240
+    },
+    {
+      "epoch": 1.7485134662469395,
+      "grad_norm": 0.2548612654209137,
+      "learning_rate": 8.879199865716479e-05,
+      "loss": 0.0903,
+      "step": 1250
+    },
+    {
+      "epoch": 1.762504372158097,
+      "grad_norm": 0.276997834444046,
+      "learning_rate": 8.717604295784657e-05,
+      "loss": 0.099,
+      "step": 1260
+    },
+    {
+      "epoch": 1.776495278069255,
+      "grad_norm": 0.27574673295021057,
+      "learning_rate": 8.556348505681147e-05,
+      "loss": 0.0935,
+      "step": 1270
+    },
+    {
+      "epoch": 1.7904861839804127,
+      "grad_norm": 0.27950429916381836,
+      "learning_rate": 8.395475221270124e-05,
+      "loss": 0.1004,
+      "step": 1280
+    },
+    {
+      "epoch": 1.8044770898915705,
+      "grad_norm": 0.3705288767814636,
+      "learning_rate": 8.235027067068175e-05,
+      "loss": 0.1005,
+      "step": 1290
+    },
+    {
+      "epoch": 1.8184679958027283,
+      "grad_norm": 0.23482586443424225,
+      "learning_rate": 8.075046554950612e-05,
+      "loss": 0.0952,
+      "step": 1300
+    },
+    {
+      "epoch": 1.832458901713886,
+      "grad_norm": 0.2208295315504074,
+      "learning_rate": 7.91557607288767e-05,
+      "loss": 0.095,
+      "step": 1310
+    },
+    {
+      "epoch": 1.8464498076250437,
+      "grad_norm": 0.3105419874191284,
+      "learning_rate": 7.756657873713513e-05,
+      "loss": 0.0817,
+      "step": 1320
+    },
+    {
+      "epoch": 1.8604407135362013,
+      "grad_norm": 0.2837795615196228,
+      "learning_rate": 7.598334063931032e-05,
+      "loss": 0.0961,
+      "step": 1330
+    },
+    {
+      "epoch": 1.8744316194473591,
+      "grad_norm": 0.25267350673675537,
+      "learning_rate": 7.440646592555442e-05,
+      "loss": 0.0954,
+      "step": 1340
+    },
+    {
+      "epoch": 1.888422525358517,
+      "grad_norm": 0.23757292330265045,
+      "learning_rate": 7.28363723999959e-05,
+      "loss": 0.0766,
+      "step": 1350
+    },
+    {
+      "epoch": 1.9024134312696748,
+      "grad_norm": 0.27169138193130493,
+      "learning_rate": 7.127347607003929e-05,
+      "loss": 0.0879,
+      "step": 1360
+    },
+    {
+      "epoch": 1.9164043371808326,
+      "grad_norm": 0.2501109540462494,
+      "learning_rate": 6.971819103614118e-05,
+      "loss": 0.0881,
+      "step": 1370
+    },
+    {
+      "epoch": 1.9303952430919902,
+      "grad_norm": 0.2925432324409485,
+      "learning_rate": 6.81709293820914e-05,
+      "loss": 0.0927,
+      "step": 1380
+    },
+    {
+      "epoch": 1.944386149003148,
+      "grad_norm": 0.29960620403289795,
+      "learning_rate": 6.663210106582838e-05,
+      "loss": 0.1028,
+      "step": 1390
+    },
+    {
+      "epoch": 1.9583770549143056,
+      "grad_norm": 0.28269627690315247,
+      "learning_rate": 6.510211381081794e-05,
+      "loss": 0.0961,
+      "step": 1400
+    },
+    {
+      "epoch": 1.9583770549143056,
+      "eval_loss": 0.10136425495147705,
+      "eval_runtime": 22.0658,
+      "eval_samples_per_second": 14.411,
+      "eval_steps_per_second": 3.626,
+      "step": 1400
+    },
+    {
+      "epoch": 1.9723679608254634,
+      "grad_norm": 0.261138379573822,
+      "learning_rate": 6.358137299802416e-05,
+      "loss": 0.0869,
+      "step": 1410
+    },
+    {
+      "epoch": 1.9863588667366212,
+      "grad_norm": 0.2551378309726715,
+      "learning_rate": 6.207028155850062e-05,
+      "loss": 0.1001,
+      "step": 1420
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.26610472798347473,
+      "learning_rate": 6.0569239866631124e-05,
+      "loss": 0.1051,
+      "step": 1430
+    },
+    {
+      "epoch": 2.013990905911158,
+      "grad_norm": 0.3179965317249298,
+      "learning_rate": 5.90786456340477e-05,
+      "loss": 0.0866,
+      "step": 1440
+    },
+    {
+      "epoch": 2.0279818118223156,
+      "grad_norm": 0.21667960286140442,
+      "learning_rate": 5.759889380425393e-05,
+      "loss": 0.0839,
+      "step": 1450
+    },
+    {
+      "epoch": 2.0419727177334734,
+      "grad_norm": 0.26058515906333923,
+      "learning_rate": 5.613037644798196e-05,
+      "loss": 0.0659,
+      "step": 1460
+    },
+    {
+      "epoch": 2.055963623644631,
+      "grad_norm": 0.2897161841392517,
+      "learning_rate": 5.467348265931063e-05,
+      "loss": 0.0785,
+      "step": 1470
+    },
+    {
+      "epoch": 2.0699545295557886,
+      "grad_norm": 0.24778111279010773,
+      "learning_rate": 5.3228598452572084e-05,
+      "loss": 0.0743,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0839454354669464,
+      "grad_norm": 0.2585526704788208,
+      "learning_rate": 5.179610666007475e-05,
+      "loss": 0.0844,
+      "step": 1490
+    },
+    {
+      "epoch": 2.0979363413781043,
+      "grad_norm": 0.27612507343292236,
+      "learning_rate": 5.037638683066903e-05,
+      "loss": 0.0978,
+      "step": 1500
+    },
+    {
+      "epoch": 2.111927247289262,
+      "grad_norm": 0.2550652325153351,
+      "learning_rate": 4.896981512918325e-05,
+      "loss": 0.0788,
+      "step": 1510
+    },
+    {
+      "epoch": 2.12591815320042,
+      "grad_norm": 0.33637735247612,
+      "learning_rate": 4.757676423675601e-05,
+      "loss": 0.0833,
+      "step": 1520
+    },
+    {
+      "epoch": 2.1399090591115772,
+      "grad_norm": 0.28613343834877014,
+      "learning_rate": 4.619760325209184e-05,
+      "loss": 0.0786,
+      "step": 1530
+    },
+    {
+      "epoch": 2.153899965022735,
+      "grad_norm": 0.22645096480846405,
+      "learning_rate": 4.483269759366569e-05,
+      "loss": 0.0726,
+      "step": 1540
+    },
+    {
+      "epoch": 2.167890870933893,
+      "grad_norm": 0.2636771500110626,
+      "learning_rate": 4.348240890290289e-05,
+      "loss": 0.0794,
+      "step": 1550
+    },
+    {
+      "epoch": 2.1818817768450507,
+      "grad_norm": 0.312181293964386,
+      "learning_rate": 4.214709494835962e-05,
+      "loss": 0.077,
+      "step": 1560
+    },
+    {
+      "epoch": 2.1958726827562085,
+      "grad_norm": 0.32286760210990906,
+      "learning_rate": 4.0827109530929595e-05,
+      "loss": 0.0798,
+      "step": 1570
+    },
+    {
+      "epoch": 2.2098635886673663,
+      "grad_norm": 0.3038155138492584,
+      "learning_rate": 3.952280239010192e-05,
+      "loss": 0.0802,
+      "step": 1580
+    },
+    {
+      "epoch": 2.223854494578524,
+      "grad_norm": 0.28648459911346436,
+      "learning_rate": 3.8234519111295365e-05,
+      "loss": 0.0833,
+      "step": 1590
+    },
+    {
+      "epoch": 2.2378454004896815,
+      "grad_norm": 0.32507434487342834,
+      "learning_rate": 3.6962601034292766e-05,
+      "loss": 0.076,
+      "step": 1600
+    },
+    {
+      "epoch": 2.2378454004896815,
+      "eval_loss": 0.09979136288166046,
+      "eval_runtime": 22.3104,
+      "eval_samples_per_second": 14.253,
+      "eval_steps_per_second": 3.586,
+      "step": 1600
+    },
+    {
+      "epoch": 2.2518363064008393,
+      "grad_norm": 0.25779277086257935,
+      "learning_rate": 3.570738516280092e-05,
+      "loss": 0.0798,
+      "step": 1610
+    },
+    {
+      "epoch": 2.265827212311997,
+      "grad_norm": 0.27955761551856995,
+      "learning_rate": 3.446920407515887e-05,
+      "loss": 0.0758,
+      "step": 1620
+    },
+    {
+      "epoch": 2.279818118223155,
+      "grad_norm": 0.3353288471698761,
+      "learning_rate": 3.3248385836218977e-05,
+      "loss": 0.0803,
+      "step": 1630
+    },
+    {
+      "epoch": 2.2938090241343128,
+      "grad_norm": 0.2533150613307953,
+      "learning_rate": 3.2045253910423864e-05,
+      "loss": 0.0754,
+      "step": 1640
+    },
+    {
+      "epoch": 2.3077999300454706,
+      "grad_norm": 0.30213606357574463,
+      "learning_rate": 3.086012707610209e-05,
+      "loss": 0.0726,
+      "step": 1650
+    },
+    {
+      "epoch": 2.3217908359566284,
+      "grad_norm": 0.3189963400363922,
+      "learning_rate": 2.9693319341005633e-05,
+      "loss": 0.0798,
+      "step": 1660
+    },
+    {
+      "epoch": 2.3357817418677858,
+      "grad_norm": 0.33093076944351196,
+      "learning_rate": 2.8545139859111446e-05,
+      "loss": 0.0719,
+      "step": 1670
+    },
+    {
+      "epoch": 2.3497726477789436,
+      "grad_norm": 0.25722455978393555,
+      "learning_rate": 2.7415892848708725e-05,
+      "loss": 0.069,
+      "step": 1680
+    },
+    {
+      "epoch": 2.3637635536901014,
+      "grad_norm": 0.30695757269859314,
+      "learning_rate": 2.630587751179422e-05,
+      "loss": 0.0822,
+      "step": 1690
+    },
+    {
+      "epoch": 2.377754459601259,
+      "grad_norm": 0.30474889278411865,
+      "learning_rate": 2.5215387954796555e-05,
+      "loss": 0.0744,
+      "step": 1700
+    },
+    {
+      "epoch": 2.391745365512417,
+      "grad_norm": 0.2297676056623459,
+      "learning_rate": 2.414471311065053e-05,
+      "loss": 0.0695,
+      "step": 1710
+    },
+    {
+      "epoch": 2.405736271423575,
+      "grad_norm": 0.4082636535167694,
+      "learning_rate": 2.3094136662242405e-05,
+      "loss": 0.0792,
+      "step": 1720
+    },
+    {
+      "epoch": 2.4197271773347326,
+      "grad_norm": 0.3242763876914978,
+      "learning_rate": 2.2063936967246136e-05,
+      "loss": 0.0823,
+      "step": 1730
+    },
+    {
+      "epoch": 2.43371808324589,
+      "grad_norm": 0.28933578729629517,
+      "learning_rate": 2.1054386984370434e-05,
+      "loss": 0.0768,
+      "step": 1740
+    },
+    {
+      "epoch": 2.447708989157048,
+      "grad_norm": 0.3418210744857788,
+      "learning_rate": 2.006575420103648e-05,
+      "loss": 0.0729,
+      "step": 1750
+    },
+    {
+      "epoch": 2.4616998950682056,
+      "grad_norm": 0.33131641149520874,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.0851,
+      "step": 1760
+    },
+    {
+      "epoch": 2.4756908009793634,
+      "grad_norm": 0.2488657683134079,
+      "learning_rate": 1.815228240247323e-05,
+      "loss": 0.0703,
+      "step": 1770
+    },
+    {
+      "epoch": 2.4896817068905213,
+      "grad_norm": 0.2939601540565491,
+      "learning_rate": 1.7227950375154912e-05,
+      "loss": 0.068,
+      "step": 1780
+    },
+    {
+      "epoch": 2.5036726128016786,
+      "grad_norm": 0.2726001441478729,
+      "learning_rate": 1.6325549388870365e-05,
+      "loss": 0.0673,
+      "step": 1790
+    },
+    {
+      "epoch": 2.517663518712837,
+      "grad_norm": 0.2877500653266907,
+      "learning_rate": 1.5445318541154942e-05,
+      "loss": 0.0722,
+      "step": 1800
+    },
+    {
+      "epoch": 2.517663518712837,
+      "eval_loss": 0.09834001958370209,
+      "eval_runtime": 21.9639,
+      "eval_samples_per_second": 14.478,
+      "eval_steps_per_second": 3.642,
+      "step": 1800
+    },
+    {
+      "epoch": 2.5316544246239943,
+      "grad_norm": 0.29231616854667664,
+      "learning_rate": 1.458749105540863e-05,
+      "loss": 0.0809,
+      "step": 1810
+    },
+    {
+      "epoch": 2.545645330535152,
+      "grad_norm": 0.3287057876586914,
+      "learning_rate": 1.3752294219102025e-05,
+      "loss": 0.0666,
+      "step": 1820
+    },
+    {
+      "epoch": 2.55963623644631,
+      "grad_norm": 0.32760322093963623,
+      "learning_rate": 1.2939949323554668e-05,
+      "loss": 0.0773,
+      "step": 1830
+    },
+    {
+      "epoch": 2.5736271423574677,
+      "grad_norm": 0.3478274643421173,
+      "learning_rate": 1.2150671605302566e-05,
+      "loss": 0.0777,
+      "step": 1840
+    },
+    {
+      "epoch": 2.5876180482686255,
+      "grad_norm": 0.29048407077789307,
+      "learning_rate": 1.1384670189069635e-05,
+      "loss": 0.078,
+      "step": 1850
+    },
+    {
+      "epoch": 2.601608954179783,
+      "grad_norm": 0.36669981479644775,
+      "learning_rate": 1.0642148032358645e-05,
+      "loss": 0.084,
+      "step": 1860
+    },
+    {
+      "epoch": 2.615599860090941,
+      "grad_norm": 0.3382461667060852,
+      "learning_rate": 9.923301871676227e-06,
+      "loss": 0.0758,
+      "step": 1870
+    },
+    {
+      "epoch": 2.6295907660020985,
+      "grad_norm": 0.3543117344379425,
+      "learning_rate": 9.228322170406056e-06,
+      "loss": 0.0665,
+      "step": 1880
+    },
+    {
+      "epoch": 2.6435816719132563,
+      "grad_norm": 0.4150249660015106,
+      "learning_rate": 8.55739306834431e-06,
+      "loss": 0.0782,
+      "step": 1890
+    },
+    {
+      "epoch": 2.657572577824414,
+      "grad_norm": 0.2844158113002777,
+      "learning_rate": 7.910692332910452e-06,
+      "loss": 0.0723,
+      "step": 1900
+    },
+    {
+      "epoch": 2.671563483735572,
+      "grad_norm": 0.39910125732421875,
+      "learning_rate": 7.288391312046583e-06,
+      "loss": 0.0731,
+      "step": 1910
+    },
+    {
+      "epoch": 2.6855543896467298,
+      "grad_norm": 0.2358187735080719,
+      "learning_rate": 6.690654888817538e-06,
+      "loss": 0.0697,
+      "step": 1920
+    },
+    {
+      "epoch": 2.699545295557887,
+      "grad_norm": 0.3523266017436981,
+      "learning_rate": 6.117641437723931e-06,
+      "loss": 0.0781,
+      "step": 1930
+    },
+    {
+      "epoch": 2.7135362014690454,
+      "grad_norm": 0.3880462348461151,
+      "learning_rate": 5.569502782739766e-06,
+      "loss": 0.0718,
+      "step": 1940
+    },
+    {
+      "epoch": 2.7275271073802028,
+      "grad_norm": 0.3129269778728485,
+      "learning_rate": 5.046384157085537e-06,
+      "loss": 0.0679,
+      "step": 1950
+    },
+    {
+      "epoch": 2.7415180132913606,
+      "grad_norm": 0.33428144454956055,
+      "learning_rate": 4.548424164747667e-06,
+      "loss": 0.069,
+      "step": 1960
+    },
+    {
+      "epoch": 2.7555089192025184,
+      "grad_norm": 0.30948323011398315,
+      "learning_rate": 4.075754743754478e-06,
+      "loss": 0.0625,
+      "step": 1970
+    },
+    {
+      "epoch": 2.769499825113676,
+      "grad_norm": 0.23879970610141754,
+      "learning_rate": 3.6285011312181873e-06,
+      "loss": 0.0801,
+      "step": 1980
+    },
+    {
+      "epoch": 2.783490731024834,
+      "grad_norm": 0.3375139832496643,
+      "learning_rate": 3.2067818301524497e-06,
+      "loss": 0.082,
+      "step": 1990
+    },
+    {
+      "epoch": 2.7974816369359914,
+      "grad_norm": 0.35570380091667175,
+      "learning_rate": 2.810708578074239e-06,
+      "loss": 0.0717,
+      "step": 2000
+    },
+    {
+      "epoch": 2.7974816369359914,
+      "eval_loss": 0.09761384129524231,
+      "eval_runtime": 21.573,
+      "eval_samples_per_second": 14.741,
+      "eval_steps_per_second": 3.708,
+      "step": 2000
+    },
+    {
+      "epoch": 2.811472542847149,
+      "grad_norm": 0.34989166259765625,
+      "learning_rate": 2.440386317398058e-06,
+      "loss": 0.0611,
+      "step": 2010
+    },
+    {
+      "epoch": 2.825463448758307,
+      "grad_norm": 0.29743126034736633,
+      "learning_rate": 2.095913167630814e-06,
+      "loss": 0.0672,
+      "step": 2020
+    },
+    {
+      "epoch": 2.839454354669465,
+      "grad_norm": 0.34374746680259705,
+      "learning_rate": 1.7773803993743177e-06,
+      "loss": 0.0769,
+      "step": 2030
+    },
+    {
+      "epoch": 2.8534452605806226,
+      "grad_norm": 0.32369616627693176,
+      "learning_rate": 1.4848724101425127e-06,
+      "loss": 0.0786,
+      "step": 2040
+    },
+    {
+      "epoch": 2.8674361664917805,
+      "grad_norm": 0.33380308747291565,
+      "learning_rate": 1.2184667019997209e-06,
+      "loss": 0.083,
+      "step": 2050
+    },
+    {
+      "epoch": 2.8814270724029383,
+      "grad_norm": 0.3711698055267334,
+      "learning_rate": 9.782338610259456e-07,
+      "loss": 0.0744,
+      "step": 2060
+    },
+    {
+      "epoch": 2.8954179783140956,
+      "grad_norm": 0.2819616198539734,
+      "learning_rate": 7.642375386145872e-07,
+      "loss": 0.0715,
+      "step": 2070
+    },
+    {
+      "epoch": 2.9094088842252535,
+      "grad_norm": 0.4502160847187042,
+      "learning_rate": 5.765344346075452e-07,
+      "loss": 0.0782,
+      "step": 2080
+    },
+    {
+      "epoch": 2.9233997901364113,
+      "grad_norm": 0.2982568144798279,
+      "learning_rate": 4.1517428227221267e-07,
+      "loss": 0.0701,
+      "step": 2090
+    },
+    {
+      "epoch": 2.937390696047569,
+      "grad_norm": 0.33230841159820557,
+      "learning_rate": 2.8019983512427293e-07,
+      "loss": 0.0778,
+      "step": 2100
+    },
+    {
+      "epoch": 2.951381601958727,
+      "grad_norm": 0.19723743200302124,
+      "learning_rate": 1.7164685559986072e-07,
+      "loss": 0.0652,
+      "step": 2110
+    },
+    {
+      "epoch": 2.9653725078698847,
+      "grad_norm": 0.31668955087661743,
+      "learning_rate": 8.954410558007497e-08,
+      "loss": 0.0709,
+      "step": 2120
+    },
+    {
+      "epoch": 2.9793634137810425,
+      "grad_norm": 0.3358490765094757,
+      "learning_rate": 3.3913338770263925e-08,
+      "loss": 0.0643,
+      "step": 2130
+    },
+    {
+      "epoch": 2.9933543196922,
+      "grad_norm": 0.29635491967201233,
+      "learning_rate": 4.769294936268765e-09,
+      "loss": 0.0747,
+      "step": 2140
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2145,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6652540184035328e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoints/checkpoint-2145/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbc9bc83f818f5138562feaa20da67402920a7f2ed95028654b04bd38ebe547e
+size 6417

checkpoints/checkpoint-2145/video_preprocessor_config.json

@@ -0,0 +1,41 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": true,
+  "device": null,
+  "do_center_crop": null,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "do_sample_frames": true,
+  "fps": 2,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "input_data_format": null,
+  "max_frames": 768,
+  "merge_size": 2,
+  "min_frames": 4,
+  "num_frames": null,
+  "pad_size": null,
+  "patch_size": 16,
+  "processor_class": "Qwen3VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "return_metadata": false,
+  "size": {
+    "longest_edge": 25165824,
+    "shortest_edge": 4096
+  },
+  "temporal_patch_size": 2,
+  "video_metadata": null,
+  "video_processor_type": "Qwen3VLVideoProcessor"
+}

dataset/dataset_info.json

@@ -0,0 +1,30 @@
+{
+  "fva_affordance_train": {
+    "file_name": "train_llamafactory.json",
+    "formatting": "sharegpt",
+    "columns": {
+      "messages": "conversations",
+      "images": "images"
+    },
+    "tags": {
+      "role_tag": "from",
+      "content_tag": "value",
+      "user_tag": "human",
+      "assistant_tag": "gpt"
+    }
+  },
+  "fva_affordance_val": {
+    "file_name": "val_llamafactory.json",
+    "formatting": "sharegpt",
+    "columns": {
+      "messages": "conversations",
+      "images": "images"
+    },
+    "tags": {
+      "role_tag": "from",
+      "content_tag": "value",
+      "user_tag": "human",
+      "assistant_tag": "gpt"
+    }
+  }
+}

inference.py

@@ -0,0 +1,101 @@
+"""
+Inference with fine-tuned Qwen3-VL-2B affordance model.
+Tests the LoRA adapter on sample images.
+"""
+
+import os
+import json
+import torch
+from PIL import Image
+from unsloth import FastVisionModel
+
+# ============================================================
+# CONFIG
+# ============================================================
+ADAPTER_PATH = "c:/Fidelity_dynamics/qwen_affordance_exp/output/lora_adapter"
+DATASET_DIR = "c:/Fidelity_dynamics/qwen_affordance_exp/dataset"
+VAL_FILE = os.path.join(DATASET_DIR, "val_msswift.jsonl")
+
+# ============================================================
+# LOAD MODEL + ADAPTER
+# ============================================================
+print("Loading model with LoRA adapter...")
+
+model, tokenizer = FastVisionModel.from_pretrained(
+    ADAPTER_PATH,
+    load_in_4bit=True,
+    max_seq_length=2048,
+)
+
+FastVisionModel.for_inference(model)
+print("Model loaded!\n")
+
+# ============================================================
+# RUN INFERENCE ON VALIDATION SAMPLES
+# ============================================================
+# Load a few val samples
+val_data = []
+with open(VAL_FILE, "r", encoding="utf-8") as f:
+    for i, line in enumerate(f):
+        if i >= 5:  # Test on first 5
+            break
+        val_data.append(json.loads(line.strip()))
+
+print("=" * 60)
+print("INFERENCE RESULTS")
+print("=" * 60)
+
+for i, sample in enumerate(val_data):
+    img_path = sample["images"][0]
+    ground_truth = sample["messages"][2]["content"]  # assistant response
+
+    # Load image
+    image = Image.open(img_path).convert("RGB")
+    image = image.resize((384, 216), Image.LANCZOS)
+
+    # Build prompt
+    messages = [
+        {
+            "role": "system",
+            "content": [{"type": "text", "text": "You are a visual affordance detection assistant. Given a first-person view image, identify what actions can be performed on the visible objects."}],
+        },
+        {
+            "role": "user",
+            "content": [
+                {"type": "image", "image": image},
+                {"type": "text", "text": "What affordances are present in this scene? Describe what actions can be performed on the visible objects."},
+            ],
+        },
+    ]
+
+    # Tokenize
+    input_text = tokenizer.apply_chat_template(
+        messages, tokenize=False, add_generation_prompt=True
+    )
+    inputs = tokenizer(
+        input_text,
+        return_tensors="pt",
+        add_special_tokens=False,
+    ).to(model.device)
+
+    # Generate
+    with torch.no_grad():
+        output_ids = model.generate(
+            **inputs,
+            max_new_tokens=256,
+            temperature=0.7,
+            do_sample=True,
+        )
+
+    # Decode only new tokens
+    new_tokens = output_ids[0][inputs["input_ids"].shape[1]:]
+    response = tokenizer.decode(new_tokens, skip_special_tokens=True)
+
+    print(f"\n--- Sample {i+1} ---")
+    print(f"Image: {os.path.basename(img_path)}")
+    print(f"Ground Truth: {ground_truth[:150]}...")
+    print(f"Prediction:   {response[:150]}...")
+    print()
+
+print("=" * 60)
+print("Inference complete!")

prepare_finetune_data.py

@@ -0,0 +1,190 @@
+import json
+import os
+import random
+
+random.seed(42)
+
+SRC = "c:/Fidelity_dynamics/qwen_affordance_exp/dataset/annotations.jsonl"
+OUT_DIR = "c:/Fidelity_dynamics/qwen_affordance_exp/dataset"
+IMG_BASE = "c:/Fidelity_dynamics/qwen_affordance_exp/dataset"
+
+# Load annotations
+annotations = []
+with open(SRC, "r") as f:
+    for line in f:
+        annotations.append(json.loads(line.strip()))
+
+print(f"Loaded {len(annotations)} annotations")
+
+# Verify all images exist
+missing = 0
+for a in annotations:
+    path = os.path.join(IMG_BASE, a['image_path'])
+    if not os.path.exists(path):
+        missing += 1
+print(f"Missing images: {missing}")
+
+# Shuffle and split: 90% train, 10% val
+random.shuffle(annotations)
+split_idx = int(len(annotations) * 0.9)
+train_data = annotations[:split_idx]
+val_data = annotations[split_idx:]
+print(f"Train: {len(train_data)}, Val: {len(val_data)}")
+
+# --- Question templates for diversity ---
+questions = [
+    "What affordance is present in this scene?",
+    "What action can be performed in this image?",
+    "Describe the affordance visible in this first-person view.",
+    "What can the person do with the objects in this scene?",
+    "Identify the possible interaction in this egocentric view.",
+]
+
+def build_answer(ann):
+    """Build a rich answer from annotation fields."""
+    affordance = ann['affordance']
+    caption = ann['raw_caption']
+    task = ann['task'].replace('_', ' ')
+    return f"The affordance in this scene is: {affordance}. The scene shows {caption}. This is a {task} task captured from an egocentric viewpoint."
+
+
+# ============================================================
+# FORMAT 1: MS-SWIFT (Official Qwen recommended)
+# ============================================================
+def to_ms_swift(data, filename):
+    out_path = os.path.join(OUT_DIR, filename)
+    with open(out_path, "w", encoding="utf-8") as f:
+        for ann in data:
+            img_path = os.path.join(IMG_BASE, ann['image_path']).replace("\\", "/")
+            question = random.choice(questions)
+            answer = build_answer(ann)
+
+            entry = {
+                "messages": [
+                    {
+                        "role": "system",
+                        "content": "You are a visual affordance detection assistant. Given a first-person view image, identify what actions can be performed on the visible objects."
+                    },
+                    {
+                        "role": "user",
+                        "content": f"<image>\n{question}"
+                    },
+                    {
+                        "role": "assistant",
+                        "content": answer
+                    }
+                ],
+                "images": [img_path]
+            }
+            f.write(json.dumps(entry, ensure_ascii=False) + "\n")
+    print(f"Written: {out_path} ({len(data)} samples)")
+
+
+# ============================================================
+# FORMAT 2: LLaMA-Factory ShareGPT
+# ============================================================
+def to_llamafactory(data, filename):
+    out_path = os.path.join(OUT_DIR, filename)
+    all_entries = []
+    for ann in data:
+        img_path = ann['image_path']  # relative path
+        question = random.choice(questions)
+        answer = build_answer(ann)
+
+        entry = {
+            "conversations": [
+                {
+                    "from": "human",
+                    "value": f"<image>\n{question}"
+                },
+                {
+                    "from": "gpt",
+                    "value": answer
+                }
+            ],
+            "images": [img_path]
+        }
+        all_entries.append(entry)
+
+    with open(out_path, "w", encoding="utf-8") as f:
+        json.dump(all_entries, f, ensure_ascii=False, indent=2)
+    print(f"Written: {out_path} ({len(data)} samples)")
+
+
+# ============================================================
+# FORMAT 3: LLaMA-Factory dataset_info.json
+# ============================================================
+def write_dataset_info():
+    info = {
+        "fva_affordance_train": {
+            "file_name": "train_llamafactory.json",
+            "formatting": "sharegpt",
+            "columns": {
+                "messages": "conversations",
+                "images": "images"
+            },
+            "tags": {
+                "role_tag": "from",
+                "content_tag": "value",
+                "user_tag": "human",
+                "assistant_tag": "gpt"
+            }
+        },
+        "fva_affordance_val": {
+            "file_name": "val_llamafactory.json",
+            "formatting": "sharegpt",
+            "columns": {
+                "messages": "conversations",
+                "images": "images"
+            },
+            "tags": {
+                "role_tag": "from",
+                "content_tag": "value",
+                "user_tag": "human",
+                "assistant_tag": "gpt"
+            }
+        }
+    }
+    out_path = os.path.join(OUT_DIR, "dataset_info.json")
+    with open(out_path, "w", encoding="utf-8") as f:
+        json.dump(info, f, indent=2)
+    print(f"Written: {out_path}")
+
+
+# Generate all formats
+print("\n" + "=" * 60)
+print("Generating MS-SWIFT format...")
+print("=" * 60)
+to_ms_swift(train_data, "train_msswift.jsonl")
+to_ms_swift(val_data, "val_msswift.jsonl")
+
+print("\n" + "=" * 60)
+print("Generating LLaMA-Factory format...")
+print("=" * 60)
+to_llamafactory(train_data, "train_llamafactory.json")
+to_llamafactory(val_data, "val_llamafactory.json")
+write_dataset_info()
+
+# Print sample from each format
+print("\n" + "=" * 60)
+print("SAMPLE: MS-SWIFT format")
+print("=" * 60)
+with open(os.path.join(OUT_DIR, "train_msswift.jsonl"), "r") as f:
+    sample = json.loads(f.readline())
+    print(json.dumps(sample, indent=2))
+
+print("\n" + "=" * 60)
+print("SAMPLE: LLaMA-Factory format")
+print("=" * 60)
+with open(os.path.join(OUT_DIR, "train_llamafactory.json"), "r") as f:
+    data = json.load(f)
+    print(json.dumps(data[0], indent=2))
+
+print("\n" + "=" * 60)
+print("ALL FILES GENERATED")
+print("=" * 60)
+for fname in os.listdir(OUT_DIR):
+    fpath = os.path.join(OUT_DIR, fname)
+    if os.path.isfile(fpath):
+        size = os.path.getsize(fpath) / 1024
+        print(f"  {fname} ({size:.1f} KB)")