LLM EXPLORER 60,062 MODELS INDEXED

Poro 34B Chat by LumiOpen

By LumiOpen · 83 downloads
HF Repo ↗ Copy URL

Poro 34B Chat is an open-source language model by LumiOpen. Features: 34b LLM, VRAM: 68GB, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.12.

  Arxiv:2404.01856   Bloom   Conversational Dataset:lumiopen/instruction-c...   En   Endpoints compatible   Fi   Instruct   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/LumiOpen/Poro-34B-chat 

Poro 34B Chat Parameters and Internals

Model Type 
Chat-tuned
Additional Notes 
Created in collaboration between SiloGen, TurkuNLP group, and HPLT.
Supported Languages 
Finnish (N/A), English (N/A)
Training Details 
Data Sources:
LumiOpen/instruction-collection-fin, databricks-dolly-15k, Curated OASST2, Argilla/10k_prompts_ranked_mistral_large_responses
Methodology:
SFT finetune on Finnish and English instruction datasets (40% English, 40% Finnish, 20% cross-lingual)
Hardware Used:
3 nodes (8 GPUs per node), LUMI supercomputer
Input Output 
Input Format:
ChatML chat template
LLM NamePoro 34B Chat
Repository 🤗https://huggingface.co/LumiOpen/Poro-34B-chat 
Model Size34b
Required VRAM68 GB
Updated2026-08-04
MaintainerLumiOpen
Model Typebloom
Instruction-BasedYes
Model Files  4.7 GB: 1-of-14   4.9 GB: 2-of-14   4.9 GB: 3-of-14   4.9 GB: 4-of-14   4.9 GB: 5-of-14   4.9 GB: 6-of-14   4.9 GB: 7-of-14   4.9 GB: 8-of-14   4.9 GB: 9-of-14   4.9 GB: 10-of-14   4.9 GB: 11-of-14   4.9 GB: 12-of-14   4.9 GB: 13-of-14   4.5 GB: 14-of-14   0.0 GB
Supported Languagesfi en
Model ArchitectureBloomForCausalLM
Licenseapache-2.0
Model Max Length2048
Transformers Version4.35.1
Tokenizer ClassBloomTokenizer
Padding Token<pad>
Vocabulary Size128000
Torch Data Typebfloat16