Instructions to use allenai/Olmo-3-7B-Think with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use allenai/Olmo-3-7B-Think with Transformers:

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

pipe = pipeline("text-generation", model="allenai/Olmo-3-7B-Think")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("allenai/Olmo-3-7B-Think")
model = AutoModelForCausalLM.from_pretrained("allenai/Olmo-3-7B-Think")
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

vLLM

How to use allenai/Olmo-3-7B-Think with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "allenai/Olmo-3-7B-Think"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "allenai/Olmo-3-7B-Think",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/allenai/Olmo-3-7B-Think

SGLang

How to use allenai/Olmo-3-7B-Think 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 "allenai/Olmo-3-7B-Think" \
    --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": "allenai/Olmo-3-7B-Think",
		"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 "allenai/Olmo-3-7B-Think" \
        --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": "allenai/Olmo-3-7B-Think",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use allenai/Olmo-3-7B-Think with Docker Model Runner:
```
docker model run hf.co/allenai/Olmo-3-7B-Think
```

Define "do_sample" explicitly in generation_config.json

by Corellios - opened Nov 28, 2025

base: refs/heads/main

←

from: refs/pr/6

Discussion Files changed

-1

Corellios

Nov 28, 2025

Hi again :)

I also got the following error when loading the model:

model = AutoModelForCausalLM.from_pretrained(MODEL_ID, dtype=torch.bfloat16, low_cpu_mem_usage=True, device_map="cuda", local_files_only=True)
Loading checkpoint shards: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:10<00:00, 3.42s/it]
The following generation flags are not valid and may be ignored: ['temperature', 'top_p']. Set TRANSFORMERS_VERBOSITY=info for more details.

When trying to save the model, it says that the default value for "do_sample" is false when not explicitly defined.

model.save_pretrained(SAVE_DIR, safe_serialization=True, max_shard_size="10GB")
2025-11-28T17:19:45.738718+0100 | get_model_compressor | INFO - skip_sparsity_compression_stats set to True. Skipping sparsity compression statistic calculations. No sparsity compressor will be applied.
Compressing model: 224it [00:05, 40.05it/s]
Traceback (most recent call last):
File "/home/thibaut/tools/llm-compressor/.venv/lib/python3.12/site-packages/transformers/generation/configuration_utils.py", line 723, in save_pretrained
self.validate(strict=True)
File "/home/thibaut/tools/llm-compressor/.venv/lib/python3.12/site-packages/transformers/generation/configuration_utils.py", line 684, in validate
raise ValueError("GenerationConfig is invalid: \n" + info_message)
ValueError: GenerationConfig is invalid:
temperature: do_sample is set to False. However, temperature is set to 0.6 -- this flag is only used in sample-based generation modes. You should set do_sample=True or unset temperature.
top_p: do_sample is set to False. However, top_p is set to 0.95 -- this flag is only used in sample-based generation modes. You should set do_sample=True or unset top_p.
If you're using a pretrained model, note that some of these attributes may be set through the model's generation_config.json file.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "", line 1, in
File "/home/thibaut/tools/llm-compressor/.venv/lib/python3.12/site-packages/llmcompressor/transformers/compression/compressed_tensors_utils.py", line 96, in save_pretrained_wrapper
original_save_pretrained.get(model, model_class)(
File "/home/thibaut/tools/llm-compressor/.venv/lib/python3.12/site-packages/transformers/modeling_utils.py", line 3930, in save_pretrained
model_to_save.generation_config.save_pretrained(save_directory)
File "/home/thibaut/tools/llm-compressor/.venv/lib/python3.12/site-packages/transformers/generation/configuration_utils.py", line 725, in save_pretrained
raise ValueError(str(exc) + "\n\nFix these issues to save the configuration.")
ValueError: GenerationConfig is invalid:
temperature: do_sample is set to False. However, temperature is set to 0.6 -- this flag is only used in sample-based generation modes. You should set do_sample=True or unset temperature.
top_p: do_sample is set to False. However, top_p is set to 0.95 -- this flag is only used in sample-based generation modes. You should set do_sample=True or unset top_p.
If you're using a pretrained model, note that some of these attributes may be set through the model's generation_config.json file.
Fix these issues to save the configuration.

Define "do_sample" explicitly in generation_config.json7ccdc12d

baileyk changed pull request status to merged Dec 10, 2025

baileyk

Ai2 org Dec 10, 2025

Hi there, thanks for the PR! Merged :)

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment