Instructions to use ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged with Transformers:

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

pipe = pipeline("text-generation", model="ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged")
model = AutoModelForMultimodalLM.from_pretrained("ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged

SGLang

How to use ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged 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 "ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged" \
    --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": "ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged",
		"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 "ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged" \
        --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": "ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged with Docker Model Runner:
```
docker model run hf.co/ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged
```

ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged

Standalone 27B merged checkpoint of ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-RealCoder-v3 (240x clean Barron-PolyStep, 20,400 steps / 3,302,400 vertex evaluations, zero train-on-test leakage) folded into Qwen/Qwen3.8-27B (bfloat16).

Merge Verification Summary (Apple M4 Max 40-Core Metal GPU)

Folded Tensors (3) (model-00017-of-00018.safetensors, lora_scale=2.0):
- model.language_model.layers.63.mlp.down_proj.weight (||W_merged - W_base||_F = 3.8712, ||W_base||_F = 109.8192)
- model.language_model.layers.63.self_attn.o_proj.weight (||W_merged - W_base||_F = 0.0001)
- model.language_model.layers.62.linear_attn.out_proj.weight (||W_merged - W_base||_F = 0.0001)
Standalone vs. PEFT Adapter Logit Cosine Similarity: 0.999987 (min 0.999973, 5/5 top-1 greedy token match)
Standalone Live Code Generation & Unit Test Execution (lora=None): 5/5 (100%)

Public Coding Leaderboard Results (`complete` / Pretrained Base Split)

BigCodeBench-Hard (148 tasks): 29.05% (43/148) vs. 27.70% (41/148) Base (+1.35%) — #1 Modern Pretrained Base Model (<= 70B)
EvalPlus HumanEval / HumanEval+ (164 tasks, 80x): 74.39% / 74.39% (122/164) vs. 71.34% / 71.34% (117/164) Base (+3.05%) — #1 Modern Pretrained Base Model (<= 70B)
EvalPlus MBPP / MBPP+ (378 tasks, 35x): 88.36% (334/378) / 75.66% (286/378) vs. 86.51% / 74.34% Base (+1.85% / +1.32%) — #1 Modern Pretrained Base Model (<= 70B)
External 40-Function Open-Ended Suite (HumanEval + MBPP + DS-1000 + AlgoGen): 97.50% (39/40) Pass@1, 98.75% (158/160) assertions (0 regressions)

Downloads last month: 2,594

Safetensors

Model size

28B params

Tensor type

BF16

Model tree for ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged

Base model

Qwen/Qwen3.8-27B

Finetuned

(816)

this model

ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged

Merge Verification Summary (Apple M4 Max 40-Core Metal GPU)

Public Coding Leaderboard Results (complete / Pretrained Base Split)

Model tree for ProbioticFarmer/Qwen3.8-27B-BarronPolyStep-Merged

Public Coding Leaderboard Results (`complete` / Pretrained Base Split)