Instructions to use hassanshka/Biomni-R0-32B-INT4-to-BF16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hassanshka/Biomni-R0-32B-INT4-to-BF16 with Transformers:

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

pipe = pipeline("text-generation", model="hassanshka/Biomni-R0-32B-INT4-to-BF16")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("hassanshka/Biomni-R0-32B-INT4-to-BF16")
model = AutoModelForCausalLM.from_pretrained("hassanshka/Biomni-R0-32B-INT4-to-BF16")
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 hassanshka/Biomni-R0-32B-INT4-to-BF16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hassanshka/Biomni-R0-32B-INT4-to-BF16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hassanshka/Biomni-R0-32B-INT4-to-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/hassanshka/Biomni-R0-32B-INT4-to-BF16

SGLang

How to use hassanshka/Biomni-R0-32B-INT4-to-BF16 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 "hassanshka/Biomni-R0-32B-INT4-to-BF16" \
    --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": "hassanshka/Biomni-R0-32B-INT4-to-BF16",
		"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 "hassanshka/Biomni-R0-32B-INT4-to-BF16" \
        --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": "hassanshka/Biomni-R0-32B-INT4-to-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use hassanshka/Biomni-R0-32B-INT4-to-BF16 with Docker Model Runner:
```
docker model run hf.co/hassanshka/Biomni-R0-32B-INT4-to-BF16
```

Biomni-R0-32B-INT4-to-BF16 (Bridge Model)

This is a dequantized BF16 version of the AWQ INT4 Biomni model. It has been converted back from INT4 → BF16 format.

Purpose

This "Bridge Model" serves several purposes:

Fine-tuning base: Use as a starting point for LoRA or full fine-tuning
Research: Study quantization/dequantization quality recovery
Compatibility: Run on hardware without INT4/FP8 support

Dequantization Details

Parameter	Value
Source	Biomni-R0-32B-AWQ-INT4-CustomCalib
Target Dtype	BFloat16
Method	Standard AWQ unpacking (W4A16)
Group Size	128

Important Notes

⚠️ This model is NOT identical to the original BF16 model.

The dequantization process recovers an approximation:

W_bf16_recovered = W_int4 × Scale
Some precision loss is expected from the quantization → dequantization roundtrip
Best used for fine-tuning where the loss can be recovered

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "hassanshka/Biomni-R0-32B-INT4-to-BF16",
    device_map="auto",
    torch_dtype=torch.bfloat16,
    trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("hassanshka/Biomni-R0-32B-INT4-to-BF16")

# Use for inference or fine-tuning

Dequantization Process

The model was dequantized using the following algorithm:

def unpack_awq_standard(packed_weight, scales):
    group_size = 128
    scales_expanded = scales.repeat_interleave(group_size, dim=1)
    
    packed_weight = packed_weight.to(torch.int32)
    unpacked_cols = []
    mask = 0xF
    
    for i in range(8):
        weight_chunk = (packed_weight >> (i * 4)) & mask
        weight_chunk = torch.where(weight_chunk >= 8, weight_chunk - 16, weight_chunk)
        unpacked_cols.append(weight_chunk)
    
    weights = torch.stack(unpacked_cols, dim=-1)
    weights = weights.view(rows, packed_cols * 8)
    
    dequantized = weights.to(torch.bfloat16) * scales_expanded.to(torch.bfloat16)
    return dequantized

License

Apache 2.0 (same as base model)

Citation

If you use this model, please cite the original Biomni model.

Downloads last month: 4

Model tree for hassanshka/Biomni-R0-32B-INT4-to-BF16

Base model

Qwen/Qwen3-32B

Finetuned

biomni/Biomni-R0-32B-Preview

Finetuned

(2)

this model