Instructions to use mtorres98/turing-rl-qwen3-8b-grpo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mtorres98/turing-rl-qwen3-8b-grpo with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-8B")
model = PeftModel.from_pretrained(base_model, "mtorres98/turing-rl-qwen3-8b-grpo")

Transformers

How to use mtorres98/turing-rl-qwen3-8b-grpo with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="mtorres98/turing-rl-qwen3-8b-grpo")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("mtorres98/turing-rl-qwen3-8b-grpo", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use mtorres98/turing-rl-qwen3-8b-grpo with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mtorres98/turing-rl-qwen3-8b-grpo"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mtorres98/turing-rl-qwen3-8b-grpo",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/mtorres98/turing-rl-qwen3-8b-grpo

SGLang

How to use mtorres98/turing-rl-qwen3-8b-grpo with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "mtorres98/turing-rl-qwen3-8b-grpo" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mtorres98/turing-rl-qwen3-8b-grpo",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "mtorres98/turing-rl-qwen3-8b-grpo" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mtorres98/turing-rl-qwen3-8b-grpo",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use mtorres98/turing-rl-qwen3-8b-grpo with Docker Model Runner:
```
docker model run hf.co/mtorres98/turing-rl-qwen3-8b-grpo
```

mtorres98 commited on 19 days ago

Commit

344cc34

verified ·

1 Parent(s): 82e8e0c

Upload qwen3_8b_job_metrics.json with huggingface_hub

Browse files

Files changed (1) hide show

qwen3_8b_job_metrics.json +39 -0

qwen3_8b_job_metrics.json ADDED Viewed

	@@ -0,0 +1,39 @@

+{
+  "paper": "Learning User Simulators with Turing Rewards",
+  "arxiv": "2606.19336",
+  "base_model": "Qwen/Qwen3-8B",
+  "paper_judge": "Qwen/Qwen3.5-397B-A17B",
+  "job_judge": "gpt-4o-mini",
+  "job_judge_reason": "OPENROUTER_API_KEY missing; using OpenAI gpt-4o-mini",
+  "gpu": "NVIDIA A100-SXM4-80GB",
+  "gpu_mem_gb": 79.25,
+  "sft_repo": "mtorres98/turing-rl-qwen3-8b-sft",
+  "grpo_repo": "mtorres98/turing-rl-qwen3-8b-grpo",
+  "sft_metrics": {
+    "train_runtime": 56.1274,
+    "train_samples_per_second": 1.71,
+    "train_steps_per_second": 0.428,
+    "total_flos": 651154773639168.0,
+    "train_loss": 0.5987233859065478,
+    "epoch": 8.0
+  },
+  "grpo_metrics": {
+    "train_runtime": 63.3782,
+    "train_samples_per_second": 0.505,
+    "train_steps_per_second": 0.126,
+    "total_flos": 0.0,
+    "train_loss": 0.0,
+    "epoch": 0.6666666666666666
+  },
+  "scale": {
+    "sft_examples": 12,
+    "sft_steps": 24,
+    "grpo_prompts": 12,
+    "grpo_steps": 8,
+    "num_generations": 4,
+    "max_completion_length": 256,
+    "lora_rank": 64,
+    "note": "Scaled LoRA slice, not the paper's 1680 GPU-hour run."
+  },
+  "finished_at": "2026-09-09T18:09:25.613925+00:00"
+}