Instructions to use mims-harvard/bio-posttrain-qwen3-4b-cpt with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mims-harvard/bio-posttrain-qwen3-4b-cpt with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="mims-harvard/bio-posttrain-qwen3-4b-cpt")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("mims-harvard/bio-posttrain-qwen3-4b-cpt")
model = AutoModelForCausalLM.from_pretrained("mims-harvard/bio-posttrain-qwen3-4b-cpt")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use mims-harvard/bio-posttrain-qwen3-4b-cpt with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mims-harvard/bio-posttrain-qwen3-4b-cpt"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mims-harvard/bio-posttrain-qwen3-4b-cpt",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/mims-harvard/bio-posttrain-qwen3-4b-cpt

SGLang

How to use mims-harvard/bio-posttrain-qwen3-4b-cpt with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "mims-harvard/bio-posttrain-qwen3-4b-cpt" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mims-harvard/bio-posttrain-qwen3-4b-cpt",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "mims-harvard/bio-posttrain-qwen3-4b-cpt" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mims-harvard/bio-posttrain-qwen3-4b-cpt",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use mims-harvard/bio-posttrain-qwen3-4b-cpt with Docker Model Runner:
```
docker model run hf.co/mims-harvard/bio-posttrain-qwen3-4b-cpt
```

Bio-posttrain Qwen3-4B CPT

Continued pre-training (CPT) checkpoint from How Post-Training Shapes Biological Reasoning Models.

Model details

Base model: Qwen/Qwen3-4B
Stage: CPT (text-only omics corpus)
Training: lr=1e-5, gradient accumulation=64, final CPT checkpoint

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("mims-harvard/bio-posttrain-qwen3-4b-cpt", trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained("mims-harvard/bio-posttrain-qwen3-4b-cpt", trust_remote_code=True)

Collection

Part of the Bio-posttrain collection on Hugging Face.

Citation

@article{bio_posttrain_2026,
  title={How Post-Training Shapes Biological Reasoning Models},
  author={...},
  year={2026}
}

Downloads last month: 12

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for mims-harvard/bio-posttrain-qwen3-4b-cpt

Base model

Qwen/Qwen3-4B-Base

Finetuned

Qwen/Qwen3-4B

Finetuned

(735)

this model

Collection including mims-harvard/bio-posttrain-qwen3-4b-cpt

bio-posttrain

Collection

Selected model checkpoints and adapters across DNA, RNA, and protein tasks from "How Post-Training Shapes Biological Reasoning Models". • 7 items • Updated 3 days ago • 2