LLM EXPLORER 62,007 MODELS INDEXED

OLMo 7B Instruct by allenai

By allenai · 475 downloads
HF Repo ↗ Copy URL

OLMo 7B Instruct is an open-source language model by allenai. Features: 7b LLM, VRAM: 13.7GB, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.22, ELO: 1073.

  Arxiv:2402.00838   Conversational   Custom code   Dataset:allenai/dolma Dataset:allenai/tulu-v2-sft-mi... Dataset:allenai/ultrafeedback ...   En   Hf olmo   Instruct   Pytorch   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/allenai/OLMo-7B-Instruct 

OLMo 7B Instruct Benchmarks

LMSys ELO 1073 vs 1272 (so35) -15.6%
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

OLMo 7B Instruct Parameters and Internals

Model Type 
Transformer style autoregressive language model
Use Cases 
Areas:
Research, Commercial applications
Primary Use Cases:
Question answering
Considerations:
Model can still generate harmful or sensitive content
Additional Notes 
This is the instruct variant, newer versions available.
Supported Languages 
English (native)
Training Details 
Data Sources:
allenai/dolma, allenai/tulu-v2-sft-mixture, allenai/ultrafeedback_binarized_cleaned
Data Volume:
Variety, details in specific dataset documentations
Methodology:
Uses SFT + DPO methods
Context Length:
2048
Model Architecture:
Based on base model OLMo with adaptations
Input Output 
Input Format:
Text
Accepted Modalities:
text
Output Format:
Text response
Performance Tips:
Use quantization for inference efficiency
LLM NameOLMo 7B Instruct
Repository 🤗https://huggingface.co/allenai/OLMo-7B-Instruct 
Model Size7b
Required VRAM13.7 GB
Updated2026-07-22
Maintainerallenai
Model Typehf_olmo
Instruction-BasedYes
Model Files  9.9 GB: 1-of-2   3.8 GB: 2-of-2   9.9 GB: 1-of-2   3.8 GB: 2-of-2
Supported Languagesen
Model ArchitectureOLMoForCausalLM
Licenseapache-2.0
Transformers Version4.35.0.dev0
Tokenizer ClassOLMoTokenizer
Padding Token<|padding|>
Vocabulary Size50280
Torch Data Typebfloat16