Text Generation
Transformers
Safetensors
qwen3_5
image-text-to-text
math
grpo
reinforcement-learning
exact-math
sgd
format-reward
conversational
Instructions to use baseten/math-sgd-qwen38_27b-expert with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use baseten/math-sgd-qwen38_27b-expert with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="baseten/math-sgd-qwen38_27b-expert") 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("baseten/math-sgd-qwen38_27b-expert") model = AutoModelForMultimodalLM.from_pretrained("baseten/math-sgd-qwen38_27b-expert", 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 baseten/math-sgd-qwen38_27b-expert with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "baseten/math-sgd-qwen38_27b-expert" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "baseten/math-sgd-qwen38_27b-expert", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/baseten/math-sgd-qwen38_27b-expert
- SGLang
How to use baseten/math-sgd-qwen38_27b-expert 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 "baseten/math-sgd-qwen38_27b-expert" \ --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": "baseten/math-sgd-qwen38_27b-expert", "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 "baseten/math-sgd-qwen38_27b-expert" \ --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": "baseten/math-sgd-qwen38_27b-expert", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use baseten/math-sgd-qwen38_27b-expert with Docker Model Runner:
docker model run hf.co/baseten/math-sgd-qwen38_27b-expert
Download assets/environment-curves/20261027/validation.csv from baseten/math-sgd-qwen38_27b-expert: direct link, hf CLI and curl.
- Browser
- Download file 6.26 kB
-
https://huggingface.co/baseten/math-sgd-qwen38_27b-expert/resolve/main/assets/environment-curves/20261027/validation.csv
- Command line
-
hf download hf://baseten/math-sgd-qwen38_27b-expert/assets/environment-curves/20261027/validation.csv
-
curl -L -o validation.csv https://huggingface.co/baseten/math-sgd-qwen38_27b-expert/resolve/main/assets/environment-curves/20261027/validation.csv
6.26 kB
| series,completed_updates,value,source_metric,complete,samples,scored,unscored,correct,truncated,completed,format_valid,format_valid_among_completed,mean_output_tokens,correctness_lower_bound,correctness_upper_bound | |
| Adam,0,0.165,validation/reward_mean,True,600,600,,99,,,,,,, | |
| Adam,5,0.18166666666666667,validation/reward_mean,True,600,600,,109,,,,,,, | |
| Adam,10,0.19,validation/reward_mean,True,600,600,,114,,,,,,, | |
| Adam,15,0.21666666666666667,validation/reward_mean,True,600,600,,130,,,,,,, | |
| Adam,20,0.25,validation/reward_mean,True,600,600,,150,,,,,,, | |
| Adam,25,0.265,validation/reward_mean,True,600,600,,159,,,,,,, | |
| Adam,30,0.2783333333333333,validation/reward_mean,True,600,600,,167,,,,,,, | |
| Adam,35,0.29833333333333334,validation/reward_mean,True,600,600,,179,,,,,,, | |
| Adam,40,0.31333333333333335,validation/reward_mean,True,600,600,,188,,,,,,, | |
| Adam,45,0.30333333333333334,validation/reward_mean,True,600,600,,182,,,,,,, | |
| Adam,50,0.32,validation/reward_mean,True,600,600,,192,,,,,,, | |
| Adam,55,0.33666666666666667,validation/reward_mean,True,600,600,,202,,,,,,, | |
| Adam,60,0.355,validation/reward_mean,True,600,600,,213,,,,,,, | |
| Adam,65,0.36666666666666664,validation/reward_mean,True,600,600,,220,,,,,,, | |
| Adam,70,0.38,validation/reward_mean,True,600,600,,228,,,,,,, | |
| Adam,75,0.38333333333333336,validation/reward_mean,True,600,600,,230,,,,,,, | |
| Adam,80,0.395,validation/reward_mean,True,600,600,,237,,,,,,, | |
| Adam,85,0.4083333333333333,validation/reward_mean,True,600,600,,245,,,,,,, | |
| Adam,90,0.425,validation/reward_mean,True,600,600,,255,,,,,,, | |
| Adam,95,0.42833333333333334,validation/reward_mean,True,600,600,,257,,,,,,, | |
| Adam,100,0.45666666666666667,validation/reward_mean,True,600,600,,274,,,,,,, | |
| Adam,105,0.485,validation/reward_mean,True,600,600,,291,,,,,,, | |
| Adam,110,0.5033333333333333,validation/reward_mean,True,600,600,,302,,,,,,, | |
| Adam,115,0.49666666666666665,validation/reward_mean,True,600,600,,298,,,,,,, | |
| Adam,120,0.4716666666666667,validation/reward_mean,True,600,600,,283,,,,,,, | |
| Adam,125,0.5416666666666666,validation/reward_mean,True,600,600,,325,,,,,,, | |
| Adam,130,0.525,validation/reward_mean,True,600,600,,315,,,,,,, | |
| Adam,135,0.5366666666666666,validation/reward_mean,True,600,600,,322,,,,,,, | |
| Adam,140,0.54,validation/reward_mean,True,600,600,,324,,,,,,, | |
| Adam,145,0.57,validation/reward_mean,True,600,600,,342,,,,,,, | |
| Adam,150,0.565,validation/reward_mean,True,600,600,,339,,,,,,, | |
| Adam,155,0.57,validation/reward_mean,True,600,600,,342,,,,,,, | |
| Adam,160,0.5883333333333334,validation/reward_mean,True,600,600,,353,,,,,,, | |
| Adam,165,0.5833333333333334,validation/reward_mean,True,600,600,,350,,,,,,, | |
| Adam,170,0.595,validation/reward_mean,True,600,600,,357,,,,,,, | |
| Adam,175,0.6283333333333333,validation/reward_mean,True,600,600,,377,,,,,,, | |
| Adam,180,0.62,validation/reward_mean,True,600,600,,372,,,,,,, | |
| Adam,185,0.625,validation/reward_mean,True,600,600,,375,,,,,,, | |
| Adam,190,0.655,validation/reward_mean,True,600,600,,393,,,,,,, | |
| Adam,195,0.6433333333333333,validation/reward_mean,True,600,600,,386,,,,,,, | |
| Adam,200,0.66,validation/reward_mean,True,600,600,,396,,,,,,, | |
| SGD + format reward,0,0.17333333333333334,eval/validation,True,600,600,0,104,329,271,132,0.4870848708487085,25873.241666666665,0.17333333333333334,0.17333333333333334 | |
| SGD + format reward,10,0.215,eval/validation,True,600,600,0,129,361,239,152,0.6359832635983264,26313.566666666666,0.215,0.215 | |
| SGD + format reward,20,0.3016666666666667,eval/validation,True,600,600,0,181,366,234,200,0.8547008547008547,25271.135,0.3016666666666667,0.3016666666666667 | |
| SGD + format reward,30,0.31,eval/validation,True,600,600,0,186,387,213,198,0.9295774647887324,25177.31,0.31,0.31 | |
| SGD + format reward,40,0.38,eval/validation,True,600,600,0,228,338,262,245,0.9351145038167938,24372.07,0.38,0.38 | |
| SGD + format reward,50,0.3933333333333333,eval/validation,True,600,600,0,236,310,290,266,0.9172413793103448,23544.863333333335,0.3933333333333333,0.3933333333333333 | |
| SGD + format reward,60,0.4083333333333333,eval/validation,True,600,600,0,245,300,300,281,0.9366666666666666,22972.725,0.4083333333333333,0.4083333333333333 | |
| SGD + format reward,70,0.435,eval/validation,True,600,600,0,261,289,311,284,0.9131832797427653,22796.366666666665,0.435,0.435 | |
| SGD + format reward,80,0.5466666666666666,eval/validation,True,600,600,0,328,216,384,349,0.9088541666666666,21417.52666666667,0.5466666666666666,0.5466666666666666 | |
| SGD + format reward,90,0.5316666666666666,eval/validation,True,600,600,0,319,223,377,344,0.9124668435013262,21325.703333333335,0.5316666666666666,0.5316666666666666 | |
| SGD + format reward,100,0.5583333333333333,eval/validation,True,600,600,0,335,224,376,351,0.9335106382978723,21334.083333333332,0.5583333333333333,0.5583333333333333 | |
| SGD + format reward,110,0.5433333333333333,eval/validation,True,600,600,0,326,220,380,351,0.9236842105263158,20928.245,0.5433333333333333,0.5433333333333333 | |
| SGD + format reward,120,0.5483333333333333,eval/validation,True,600,600,0,329,212,388,357,0.9201030927835051,20669.585,0.5483333333333333,0.5483333333333333 | |
| SGD + format reward,130,0.58,eval/validation,True,600,600,0,348,203,397,370,0.9319899244332494,20450.351666666666,0.58,0.58 | |
| SGD + format reward,140,0.5766666666666667,eval/validation,True,600,600,0,346,216,384,366,0.953125,20709.033333333333,0.5766666666666667,0.5766666666666667 | |
| SGD + format reward,150,0.57,eval/validation,True,600,600,0,342,201,399,366,0.9172932330827067,20427.183333333334,0.57,0.57 | |
| SGD + format reward,160,0.5466666666666666,eval/validation,True,600,600,0,328,205,395,353,0.8936708860759494,20703.773333333334,0.5466666666666666,0.5466666666666666 | |
| SGD + format reward,170,0.5416666666666666,eval/validation,True,600,600,0,325,140,460,425,0.9239130434782609,19961.003333333334,0.5416666666666666,0.5416666666666666 | |
| SGD + format reward,180,0.48580968280467446,eval/validation,False,600,599,1,291,82,518,483,0.9324324324324325,18022.326666666668,0.485,0.4866666666666667 | |
| SGD + format reward,190,0.5216666666666666,eval/validation,True,600,600,0,313,66,534,505,0.9456928838951311,17293.948333333334,0.5216666666666666,0.5216666666666666 | |
| SGD + format reward,200,0.48833333333333334,eval/validation,True,600,600,0,293,29,571,528,0.9246935201401051,14941.75,0.48833333333333334,0.48833333333333334 | |