Update README for cleaned repo

README.md

@@ -1,28 +1,30 @@
 ---
 tags:
-  - training-scripts
-  - fine-tuning
-  - lora
+- fine-tuning
+- lora
+- humaneval
+- qwen3
 license: apache-2.0
 ---
 
 # Training Scripts
 
-Reusable training scripts for HF Jobs.
+Fine-tuning script for reproducing the HuggingFace fine-tuning challenge.
 
-## Scripts
+## Script
 
-| Script | Purpose |
-|--------|---------|
-| `train_sft_demo.py` | SFT on Capybara dataset |
-| `train_concise.py` | SFT on concise-tech-explanations |
+**train_humaneval_clean.py** - Fine-tune Qwen3-0.6B to beat base on HumanEval.
 
 ## Usage
 
 ```bash
-hf jobs uv run --flavor t4-small --timeout 30m --secrets HF_TOKEN   "https://huggingface.co/passagereptile455/training-scripts/resolve/main/train_sft_demo.py"
+hf jobs uv run   --flavor a10g-small   --timeout 14400   --secrets HF_TOKEN   "https://huggingface.co/passagereptile455/training-scripts/resolve/main/train_humaneval_clean.py"
 ```
 
+## Results
+
+See [qwen3-humaneval-sft](https://huggingface.co/passagereptile455/qwen3-humaneval-sft) for the trained model and full results.
+
 ## Notes
 
-Scripts use PEP 723 inline dependencies — no local setup required.
+Script uses PEP 723 inline dependencies - no local setup required.