LLM EXPLORER
 LLM Explorer: A Curated Large Language Model Directory and Analytics
Dark Theme
  LLM  List    LLM Hosting    LLM Leaderboards    Blog    Newsfeed    Advertise

Bielik 7B V0.1 by speakleash

 »  All LLMs  »  speakleash  »  Bielik 7B V0.1   URL Share it on

  3992   
  Arxiv:2410.18565   Autotrain compatible   Continuously pretrained   Endpoints compatible   Mistral   Pl   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/speakleash/Bielik-7B-v0.1 

Bielik 7B V0.1 Benchmarks

ARC: 45.22 vs 96.7 (so35)-53.2%
HellaSwag: 67.92 vs 95.3 (gpt4)-28.7%
MMLU: 47.16 vs 88.3 (so35)-46.6%
TruthfulQA: 43.2 vs 59 (gpt4)-26.8%
WinoGrande: 66.85 vs 87.5 (gpt4)-23.6%
GSM8K: 29.49 vs 96.4 (so35)-69.4%
LLME Score: 0.27088
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Bielik 7B V0.1 (speakleash/Bielik-7B-v0.1)

Bielik 7B V0.1 Parameters and Internals

Model Type 
causal decoder-only
Use Cases 
Areas:
research, commercial applications
Limitations:
Not intended for deployment without fine-tuning, Not for human-facing interactions without guardrails, Can produce factually incorrect outputs
Additional Notes 
This is a base model for further fine-tuning. A demo and chat arena are available for evaluation.
Supported Languages 
language (Polish), proficiency (high)
Training Details 
Data Sources:
Polish text corpora, SpeakLeash project
Data Volume:
36 billion tokens
Methodology:
Two epochs over text corpora
Context Length:
4096
Hardware Used:
256 NVidia GH200 cards
Model Architecture:
Similar to LLaMA and Mistral
Input Output 
Input Format:
text input
Accepted Modalities:
text
Output Format:
text
Performance Tips:
Use smaller precision (bfloat16) for less memory usage
Release Notes 
Version:
v0.1
Notes:
Base model for Polish language processing trained with SpeakLeash data, can be fine-tuned for various applications.
LLM NameBielik 7B V0.1
Repository 🤗https://huggingface.co/speakleash/Bielik-7B-v0.1 
Model Size7b
Required VRAM14.4 GB
Updated2024-12-21
Maintainerspeakleash
Model Typemistral
Model Files  4.9 GB: 1-of-3   5.0 GB: 2-of-3   4.5 GB: 3-of-3
Supported Languagespl
Model ArchitectureMistralForCausalLM
Licenseapache-2.0
Context Length8192
Model Max Length8192
Transformers Version4.37.2
Tokenizer ClassLlamaTokenizer
Vocabulary Size32000
Torch Data Typebfloat16

Best Alternatives to Bielik 7B V0.1

Best Alternatives
Context / RAM
Downloads
Likes
...Nemo Instruct 2407 Abliterated1000K / 24.5 GB26999
MegaBeam Mistral 7B 512K512K / 14.4 GB715646
SpydazWeb AI HumanAI RP512K / 14.4 GB331
SpydazWeb AI HumanAI 002512K / 14.4 GB201
...daz Web AI ChatML 512K Project512K / 14.5 GB120
MegaBeam Mistral 7B 300K282K / 14.4 GB330815
Hebrew Mistral 7B 200K256K / 30 GB327015
Astral 256K 7B V2250K / 14.4 GB190
Astral 256K 7B250K / 14.4 GB160
Boptruth Agatha 7B128K / 14.4 GB4980
Note: green Score (e.g. "73.2") means that the model is better than speakleash/Bielik-7B-v0.1.

Rank the Bielik 7B V0.1 Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

Instruction Following and Task Automation  
Factuality and Completeness of Knowledge  
Censorship and Alignment  
Data Analysis and Insight Generation  
Text Generation  
Text Summarization and Feature Extraction  
Code Generation  
Multi-Language Support and Translation  

What open-source LLMs or SLMs are you in search of? 40013 in total.

Email us: info@extractum.io. Our Privacy Policy  | Terms and Conditions  | Suggest an improvement.
Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241217
  Support LLM Explorer