# PlasmidGPT-GRPO

A GRPO-trained causal language model for plasmid/DNA sequence generation.

This update replaces the model weights with a newer checkpoint while keeping the tokenizer and configs unchanged.

## Weights
- File: `model.safetensors` (checkpoint-2300)
- Other files (config and tokenizer artifacts) are unchanged.

## Training Run
- Weights and metrics reference: https://wandb.ai/ucl-cssb/PlasmidRL/runs/ty13u43j/overview

## Usage
Install dependencies:
```
pip install torch transformers safetensors
```

Load with Transformers:
```
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "UCL-CSSB/PlasmidGPT-GRPO"
tokenizer = AutoTokenizer.from_pretrained(model_id)
if tokenizer.pad_token is None:
    tokenizer.pad_token = tokenizer.eos_token
model = AutoModelForCausalLM.from_pretrained(model_id)

prompt = "ATG"
inputs = tokenizer([prompt], return_tensors="pt")
outputs = model.generate(
    **inputs,
    max_new_tokens=128,
    do_sample=True,
    temperature=0.7,
    top_p=0.9,
    pad_token_id=tokenizer.eos_token_id,
    eos_token_id=tokenizer.eos_token_id,
)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

Notes:
- Use sampling (`temperature`, `top_p`) for diverse sequences; disable sampling for deterministic outputs.
- Runs on CPU, CUDA, or Apple MPS depending on your PyTorch installation.

---
Changelog:
- Update weights to checkpoint-2300
- Simplify README and add W&B run link