Upload 8 files

README(1).md

@@ -0,0 +1,76 @@
+---
+datasets:
+- ray0rf1re/FineWeb-Nano
+- Nix-ai/Cat-v2.8
+- databricks/databricks-dolly-15k
+- allenai/ai2_arc
+- lighteval/MATH-Hard
+- tatsu-lab/alpaca
+- HuggingFaceTB/smoltalk
+- openai/gsm8k
+language:
+- en
+license: other
+license_name: hyper-v2
+tags:
+- ai-evaluation
+- mathematics
+- trigonometry
+- calculus
+- algebra
+- qwen2
+- causal-lm
+- trained-from-scratch
+base_model: none
+---
+
+# HyperNix.2
+
+**Version:** 0.2  
+**Parameters:** 101,370,880 (~101.37 M)  
+**Architecture:** Qwen 2.5-style decoder-only transformer (trained **from scratch**)  
+
+## What can it do?
+
+| Domain | Capability |
+|---|---|
+| AI Evaluation | Test, evaluate, grade and rate other AI model outputs |
+| Mathematics | Full Trigonometry, Calculus (limits/derivatives/integrals), Algebra 1 & 2 |
+| English | Fluent conversational English |
+
+## Architecture
+
+| Hyperparameter | Value |
+|---|---|
+| `vocab_size` | 151,936 (Qwen 2.5 tokenizer) |
+| `hidden_size` | 512 |
+| `intermediate_size` | 1,193 (SwiGLU) |
+| `num_hidden_layers` | 9 |
+| `num_attention_heads` | 8 |
+| `num_key_value_heads` | 4 (GQA) |
+| `max_position_embeddings` | 2,048 |
+| `tie_word_embeddings` | True |
+| Total parameters | **101,370,880** |
+
+## Training
+
+- Hardware: Single NVIDIA GTX 1080 (8 GB VRAM)
+- Precision: FP16 + gradient checkpointing
+- Optimizer: AdamW (lr=3e-4, cosine decay)
+- This is **not** a LoRA or fine-tune — all weights are randomly initialised and trained from scratch.
+
+## Usage
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+tokenizer = AutoTokenizer.from_pretrained("ray0rf1re/hyper-Nix.2")
+model     = AutoModelForCausalLM.from_pretrained("ray0rf1re/hyper-Nix.2", torch_dtype=torch.float16)
+
+prompt = "<|im_start|>user\nEvaluate this AI response: The capital of France is London.<|im_end|>\n<|im_start|>assistant\n"
+inputs = tokenizer(prompt, return_tensors="pt")
+with torch.no_grad():
+    out = model.generate(**inputs, max_new_tokens=200, temperature=0.7)
+print(tokenizer.decode(out[0], skip_special_tokens=False))
+```

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3de485dbfafa5c3abed8c5f81a9f64ab8f067f2e321454c935f9a2c0d90a28e
+size 405532464

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892

tokenizer_config.json

@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_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|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}