Instructions to use feyninc/sqrl-35b-a3b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use feyninc/sqrl-35b-a3b with Transformers:

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

pipe = pipeline("text-generation", model="feyninc/sqrl-35b-a3b")
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)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("feyninc/sqrl-35b-a3b")
model = AutoModelForMultimodalLM.from_pretrained("feyninc/sqrl-35b-a3b", 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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use feyninc/sqrl-35b-a3b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "feyninc/sqrl-35b-a3b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "feyninc/sqrl-35b-a3b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/feyninc/sqrl-35b-a3b

SGLang

How to use feyninc/sqrl-35b-a3b 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 "feyninc/sqrl-35b-a3b" \
    --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": "feyninc/sqrl-35b-a3b",
		"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 "feyninc/sqrl-35b-a3b" \
        --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": "feyninc/sqrl-35b-a3b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use feyninc/sqrl-35b-a3b with Docker Model Runner:
```
docker model run hf.co/feyninc/sqrl-35b-a3b
```

bhavnicksm commited on Jul 7

Commit

d2f1c62

verified ·

1 Parent(s): 6b10a13

Model card: measured pass@1 and vote-of-8 results, family comparison, voting notes

Browse files

Files changed (1) hide show

README.md +22 -5

README.md CHANGED Viewed

@@ -33,13 +33,30 @@ a 3-model LLM-judge panel; BIRD-dev was never trained on.
 ## Results
 | Benchmark | Metric | Score |
 |---|---|---|
-| BIRD-dev (full 1534) | pass@1, in-training eval | **70.3%** EX |
-| BIRD-dev (clean-episode harness re-eval) | pass@1 | ~68% EX |
-Estimated pass@8 oracle ~80-82% (extrapolated from the 4B/9B oracle gaps; not yet
-measured on this checkpoint). The distilled 9B reaches 66.9% and the 4B 65.4% pass@1.
 ## How it works — the agentic protocol

 ## Results
+All BIRD-dev numbers measured on the full 1534-question set at temperature 0.7 with the
+agentic harness (execution-clustered majority vote = sample 8 candidates, group them by
+identical execution result set, return the largest cluster's query).
 | Benchmark | Metric | Score |
 |---|---|---|
+| BIRD-dev | pass@1 (avg over 8 samples) | **68.7%** EX |
+| BIRD-dev | vote-of-8 (execution-clustered) | **70.6%** EX |
+Family comparison on identical seeds (pass@1 / vote@8):
+[sqrl-4b](https://huggingface.co/feyninc/sqrl-4b) 64.6 / 68.8 ·
+[sqrl-9b](https://huggingface.co/feyninc/sqrl-9b) 66.6 / 69.8 ·
+[sqrl-35b-a3b](https://huggingface.co/feyninc/sqrl-35b-a3b) 68.7 / 70.6.
+Notes from the measurement campaign:
+- **70.6% vote-of-8 is the family's best system number**, but only +0.8 over the 9B's —
+  if serving cost matters, the distilled [sqrl-9b](https://huggingface.co/feyninc/sqrl-9b)
+  with voting is the better deployment.
+- This model is heavily RL-sharpened: it answers unanimously on 1,072/1534 questions
+  (all 8 samples produce the same result). That gives it the family's best pass@1 but
+  the smallest voting gain (+1.9) — sharpening traded sample diversity for single-shot
+  reliability.
+- Vote unanimity is a strong free confidence signal: unanimous answers are right ~87% of
+  the time; narrow/tied votes far less — useful for routing or escalation.
 ## How it works — the agentic protocol