Meta Llama 3.1 8B Instruct Bnb 4bit by unsloth

 ยป  All LLMs  ยป  unsloth  ยป  Meta Llama 3.1 8B Instruct Bnb 4bit   URL Share it on

  Arxiv:2204.05149   4-bit   4bit   Autotrain compatible Base model:meta-llama/llama-3.... Base model:quantized:meta-llam...   Bitsandbytes   Conversational   En   Endpoints compatible   Facebook   Instruct   Llama   Llama-3   Meta   Quantized   Region:us   Safetensors   Unsloth

Meta Llama 3.1 8B Instruct Bnb 4bit Benchmarks

nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Meta Llama 3.1 8B Instruct Bnb 4bit (unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit)

Meta Llama 3.1 8B Instruct Bnb 4bit Parameters and Internals

Model Type 
auto-regressive, transformer, instruction tuned, multilingual, text generation
Use Cases 
Areas:
Commercial use, Research use, Education, Climate, Open innovation
Applications:
Chatbots, Synthetic data generation and distillation, Natural language generation tasks
Primary Use Cases:
Assistant-like chat, Multilingual dialogue applications
Limitations:
Use in unsupported languages without further tuning, Violation of laws, Out-of-scope uses as per Acceptable Use Policy
Considerations:
Comprehensive safety and performance testing required before deployment
Additional Notes 
Model supports a longer context window and leverages Grouped-Query Attention for enhanced inference.
Supported Languages 
English (fluent), German (fluent), French (fluent), Italian (fluent), Portuguese (fluent), Hindi (fluent), Spanish (fluent), Thai (fluent)
Training Details 
Data Sources:
publicly available online data
Data Volume:
15 trillion tokens
Methodology:
Supervised fine-tuning (SFT) and Reinforcement Learning with Human Feedback (RLHF)
Context Length:
128000
Training Time:
1.46M GPU hours for 8B
Hardware Used:
H100-80GB GPUs
Model Architecture:
Optimized transformer architecture using Grouped-Query Attention (GQA)
Safety Evaluation 
Methodologies:
red-teaming, adversarial testing
Findings:
to be determined
Risk Categories:
misinformation, bias
Ethical Considerations:
Safety fine-tuning, safety datasets usage
Responsible Ai Considerations 
Fairness:
Efforts were made to ensure fairness across languages and tasks
Transparency:
Model capabilities and limitations are communicated openly
Accountability:
Model provided with usage guides and responsible deployment methods
Mitigation Strategies:
Safety datasets, red-teaming, responsible use guides
Input Output 
Input Format:
Multilingual text
Accepted Modalities:
text
Output Format:
Text and code
Performance Tips:
Ensure updated software and compatible hardware
Release Notes 
Version:
3.1
Date:
July 23, 2024
Notes:
First release with multilingual support and instruction fine-tuning.
LLM NameMeta Llama 3.1 8B Instruct Bnb 4bit
Repository ๐Ÿค—https://huggingface.co/unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit 
Base Model(s)  meta-llama/Llama-3.1-8B-Instruct   meta-llama/Llama-3.1-8B-Instruct
Model Size8b
Required VRAM5.7 GB
Updated2025-02-15
Maintainerunsloth
Model Typellama
Instruction-BasedYes
Model Files  5.7 GB
Supported Languagesen
Quantization Type4bit
Model ArchitectureLlamaForCausalLM
Licensemeta
Context Length131072
Model Max Length131072
Transformers Version4.44.2
Tokenizer ClassPreTrainedTokenizerFast
Padding Token<|finetune_right_pad_id|>
Vocabulary Size128256
Torch Data Typebfloat16

Quantized Models of the Meta Llama 3.1 8B Instruct Bnb 4bit

Model
Likes
Downloads
VRAM
...ball Alpaca Llama3.1 8B Philos1016 GB
Loramerged08621 GB
...ardian V0.1 13Oct2024 Epoch2.01361716 GB
...leverBoi Llama 3.1 8B Instruct13520 GB
Llama Uft1.2 Persianqa01615 GB
Finetuned8b0616 GB
Llama Uft7 Persianqa0635 GB
... 8B Instruct PRM800K Reasoning11316 GB
Merged LLama8b Abstracts0620 GB
Merged LLama8b Books0560 GB

Best Alternatives to Meta Llama 3.1 8B Instruct Bnb 4bit

Best Alternatives
Context / RAM
Downloads
Likes
...B Instruct Gradient 1048K 4bit1024K / 4.5 GB352
...B Instruct Gradient 1048K 8bit1024K / 8.6 GB161
...truct Gradient 1048K Bpw6 EXL21024K / 6.7 GB152
...truct Gradient 1048K Bpw5 EXL21024K / 5.8 GB100
Llama 3 8B Instruct 1048K 4bit1024K / 4.5 GB1525
Llama 3 8B Instruct 1048K 8bit1024K / 8.6 GB2917
... Gradient 1048K 8.0bpw H8 EXL21024K / 8.6 GB123
...ct Gradient 1048K Bpw2.25 EXL21024K / 3.4 GB91
Llama 3 8B Instruct 262K 2bit256K / 2.5 GB111
...B Instruct 262k V2 EXL2 5.0bpw256K / 5.8 GB91

Rank the Meta Llama 3.1 8B Instruct Bnb 4bit Capabilities

๐Ÿ†˜ 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! ๐ŸŒŸ

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? 43137 in total.

Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241227