LLM EXPLORER 59,694 MODELS INDEXED

LongForm OPT 350M by akoksal

By akoksal · 9 downloads
HF Repo ↗ Copy URL

LongForm OPT 350M is an open-source language model by akoksal. Features: 350m LLM, VRAM: 1.3GB, Context: 2K, Instruction-Based, LLM Explorer Score: 0.07.

  Arxiv:2304.08460   Dataset:akoksal/longform   De   En   Endpoints compatible   Es   Fr   Instruct   It   Multilingual   Opt   Pytorch   Region:us
Model Card on HF 🤗: https://huggingface.co/akoksal/LongForm-OPT-350M 

LongForm OPT 350M Parameters and Internals

Model Type 
instruction-tuning, text-generation-inference, text2text-generation
Additional Notes 
Focuses on long text generation; may have hallucination issues.
Supported Languages 
en (full_proficiency), multilingual (partial_proficiency), de (partial_proficiency), it (partial_proficiency), es (partial_proficiency), fr (partial_proficiency)
Training Details 
Data Sources:
C4, Wikipedia, Stack Exchange, WikiHow
Methodology:
reverse instructions
Input Output 
Input Format:
Prompts with 'EOI' to indicate end of instruction
Accepted Modalities:
text
Output Format:
Text outputs
LLM NameLongForm OPT 350M
Repository 🤗https://huggingface.co/akoksal/LongForm-OPT-350M 
Model Size350m
Required VRAM1.3 GB
Updated2026-07-14
Maintainerakoksal
Model Typeopt
Instruction-BasedYes
Model Files  1.3 GB
Supported Languagesen de it es fr
Model ArchitectureOPTForCausalLM
Context Length2048
Model Max Length2048
Transformers Version4.28.0.dev0
Tokenizer ClassGPT2Tokenizer
Beginning of Sentence Token</s>
End of Sentence Token</s>
Unk Token</s>
Vocabulary Size50272
Torch Data Typefloat32
Activation Functionrelu
Errorsreplace

Best Alternatives to LongForm OPT 350M

Best Alternatives
Context / RAM
Downloads
Likes
Aira OPT 350M2K / 0 GB90
Opt 350M Instruct2K / 1.3 GB174
...speed Chat Step2 Model Opt350m2K / 0.7 GB41
Chat Opt 350M Reward Deepspeed2K / 0.7 GB177
Note: green Score (e.g. "73.2") means that the model is better than akoksal/LongForm-OPT-350M.