MedGPT-OSS-20B

Merged LoRA model based on openai/gpt-oss-20b.

Model Details

• Base Model: openai/gpt-oss-20b
• Languages: en
• Type: Merged LoRA (adapter weights merged into base)
• Library: 🤗 Transformers

LoRA Configuration

• r: 8
• alpha (lora_alpha): 16
• dropout: 0.0
• target_modules: k_proj, o_proj, v_proj, q_proj
• bias: none

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_name = "justinj92/MedGPT-OSS-20B"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float16,
    device_map="auto"
)

prompt = "Your prompt here"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=128)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Merge Process

Base model and LoRA adapter weights were merged using PeftModel.merge_and_unload() to create a standalone model (no PEFT dependency required at inference).

Limitations & Bias

This model inherits the limitations and potential biases of openai/gpt-oss-20b and the fine-tuning dataset.

License

Licensed under apache-2.0. Check the base model's license for additional terms.