huggingface--transformers/docs/source/en/model_doc/flex_olmo.md

This model was published in HF papers on 2025-07-09 and contributed to Hugging Face Transformers on 2025-09-18.

FlexOlmo

FlexOlmo is a new class of language models (LMs) that supports (1) distributed training without data sharing, where different model parameters are independently trained on closed datasets, and (2) data-flexible inference, where these parameters along with their associated data can be flexibly included or excluded from model inferences with no further training. FlexOlmo employs a mixture-of-experts (MoE) architecture where each expert is trained independently on closed datasets and later integrated through a new domain-informed routing without any joint training. FlexOlmo is trained on FlexMix, a corpus we curate comprising publicly available datasets alongside seven domain-specific sets, representing realistic approximations of closed sets.

You can find all the original FlexOlmo checkpoints under the FlexOlmo collection.

Tip

Click on the FlexOlmo models in the right sidebar for more examples of how to apply FlexOlmo to different language tasks.

The example below demonstrates how to generate text with [Pipeline], [AutoModel] and from the command line.

from transformers import pipeline


pipe = pipeline(
    task="text-generation",
    model="allenai/FlexOlmo-7x7B-1T",
    device=0,
)

result = pipe("Plants create energy through a process known as")
print(result)

from transformers import AutoModelForCausalLM, AutoTokenizer


tokenizer = AutoTokenizer.from_pretrained(
    "allenai/FlexOlmo-7x7B-1T"
)

model = AutoModelForCausalLM.from_pretrained(
    "allenai/FlexOlmo-7x7B-1T",
    device_map="auto",
    attn_implementation="sdpa"
)
input_ids = tokenizer("Plants create energy through a process known as", return_tensors="pt").to(model.device)

output = model.generate(**input_ids, max_length=50, cache_implementation="static")
print(tokenizer.decode(output[0], skip_special_tokens=True))

Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the Quantization overview for more available quantization backends.

The example below uses torchao to only quantize the weights to 4-bits.

#pip install torchao
from transformers import AutoModelForCausalLM, AutoTokenizer, TorchAoConfig


torchao_config = TorchAoConfig(
    "int4_weight_only",
    group_size=128
)

tokenizer = AutoTokenizer.from_pretrained(
    "allenai/FlexOlmo-7x7B-1T"
)

model = AutoModelForCausalLM.from_pretrained(
    "allenai/FlexOlmo-7x7B-1T",
    quantization_config=torchao_config,
    device_map="auto",
    attn_implementation="sdpa"
)
input_ids = tokenizer("Plants create energy through a process known as", return_tensors="pt").to(model.device)

output = model.generate(**input_ids, max_length=50, cache_implementation="static")
print(tokenizer.decode(output[0], skip_special_tokens=True))

FlexOlmoConfig

autodoc FlexOlmoConfig

FlexOlmoForCausalLM

autodoc FlexOlmoForCausalLM

FlexOlmoModel

autodoc FlexOlmoModel - forward

FlexOlmoPreTrainedModel

autodoc FlexOlmoPreTrainedModel - forward