📖 Simple-Stories-Hindi-10M-GGUF (FP16)

This repository contains the FP16 GGUF model file for SmallScale/Simple-Stories-Hindi-10M, a 11.45M parameter decoder-only Transformer model trained from scratch on the SmallScale/Simple-Stories-Hindi dataset (~2.11M Hindi stories).

---

📌 Model Information & Base Model

• Base Model: SmallScale/Simple-Stories-Hindi-10M
• Dataset Used: SmallScale/Simple-Stories-Hindi (~2.11M simple Hindi stories)
• Best Validation Loss: 1.8157 (at step 202,000)

---

📁 Model Files

File Name	Quantization	Size	Description
Simple-Stories-Hindi-11M-F16.gguf	F16 (16-bit Float)	~22.0 MB	Full 16-bit float precision GGUF format

---

🏗️ Model Architecture Details

Parameter	Value
Parameters	11.45M total (10.17M non-embedding)
Architecture	LLaMA-style (RoPE + SwiGLU + RMSNorm)
Context Length	512 tokens
Embedding Dim (d_model)	320
FFN Hidden Dim	896
Layers	7
Attention Heads	5
Vocabulary	4,000 (SentencePiece Unigram)

---

🚀 Quick Start / Usage

1. Using with llama.cpp CLI

./llama-cli -m Simple-Stories-Hindi-11M-F16.gguf \
    -p "एक समय की बात है" \
    -n 150 \
    --temp 0.8 \
    --top-k 40 \
    --top-p 0.95

2. Using with llama-cpp-python

from llama_cpp import Llama

llm = Llama(
    model_path="Simple-Stories-Hindi-11M-F16.gguf",
    n_ctx=512,
    verbose=False
)

output = llm(
    "एक समय की बात है",
    max_tokens=150,
    temperature=0.8,
    top_k=40,
    top_p=0.95
)

print(output["choices"][0]["text"])

---

📄 License

MIT License