LLM EXPLORER 60,277 MODELS INDEXED

Gradientai Llama 3 8B Instruct 262K V2 EXL2 5.0bpw by bullerwins

By bullerwins · 5 downloads
HF Repo ↗ Copy URL

Gradientai Llama 3 8B Instruct 262K V2 EXL2 5.0bpw is an open-source language model by bullerwins. Features: 8b LLM, VRAM: 5.8GB, Context: 256K, License: llama3, Quantized, Instruction-Based, LLM Explorer Score: 0.12.

  Arxiv:2305.14233   Arxiv:2309.00071   Arxiv:2402.08268   5-bit   Conversational   En   Endpoints compatible   Exl2   Instruct   Llama   Llama-3   Meta   Quantized   Region:us   Safetensors
Model Card on HF 🤗: https://huggingface.co/bullerwins/gradientai_Llama-3-8B-Instruct-262k_v2_exl2_5.0bpw 

Gradientai Llama 3 8B Instruct 262K V2 EXL2 5.0bpw Parameters and Internals

Model Type 
text generation, instruction-tuned
Use Cases 
Areas:
commercial, research
Applications:
assistant-like chat
Primary Use Cases:
natural language generation tasks
Limitations:
English only; out-of-scope for illegal or prohibited use cases
Considerations:
Developers may fine-tune for languages beyond English adhering to specific policies.
Additional Notes 
The model is intended for English-language applications and could be adapted for other languages.
Supported Languages 
languages_supported (English), proficiency ()
Training Details 
Data Sources:
SlimPajama, UltraChat
Data Volume:
15 trillion tokens
Methodology:
Supervised fine-tuning, reinforcement learning with human feedback (RLHF)
Context Length:
262144
Hardware Used:
Crusoe Energy high performance L40S cluster
Model Architecture:
Transformer with improved RoPE theta, NTK-aware interpolation
Safety Evaluation 
Methodologies:
red teaming, adversarial evaluations
Risk Categories:
misinformation, bias, cybersecurity, child safety
Ethical Considerations:
Iterative testing for CBRNE threats
Responsible Ai Considerations 
Fairness:
Model is optimized for safety and helpfulness but trade-offs exist.
Transparency:
Steps for safety best practices outlined in Responsible Use Guide.
Accountability:
Meta
Mitigation Strategies:
Meta Llama Guard 2 and Code Shield provided for safety tailored applications
Input Output 
Input Format:
text input
Accepted Modalities:
text
Output Format:
text generation
Release Notes 
Version:
5/3/2024
Date:
2024-05-03
Notes:
Further fine-tuned for assistant-like chat ability; extended context length.
LLM NameGradientai Llama 3 8B Instruct 262k V2 EXL2 5.0bpw
Repository 🤗https://huggingface.co/bullerwins/gradientai_Llama-3-8B-Instruct-262k_v2_exl2_5.0bpw 
Model Size8b
Required VRAM5.8 GB
Updated2026-07-19
Maintainerbullerwins
Model Typellama
Instruction-BasedYes
Model Files  5.8 GB
Supported Languagesen
Quantization Typeexl2
Model ArchitectureLlamaForCausalLM
Licensellama3
Context Length262144
Model Max Length262144
Transformers Version4.41.0.dev0
Tokenizer ClassPreTrainedTokenizerFast
Vocabulary Size128256
Torch Data Typebfloat16

Best Alternatives to Gradientai Llama 3 8B Instruct 262K V2 EXL2 5.0bpw

Best Alternatives
Context / RAM
Downloads
Likes
...B Instruct Gradient 1048K 4bit1024K / 4.5 GB102
...B Instruct Gradient 1048K 8bit1024K / 8.6 GB111
...truct Gradient 1048K Bpw6 EXL21024K / 6.7 GB22
...truct Gradient 1048K Bpw5 EXL21024K / 5.8 GB50
Llama 3 8B Instruct 1048K 4bit1024K / 4.5 GB52525
Llama 3 8B Instruct 1048K 8bit1024K / 8.6 GB21317
... Gradient 1048K 8.0bpw H8 EXL21024K / 8.6 GB23
...ct Gradient 1048K Bpw2.25 EXL21024K / 3.4 GB51
Llama 3 8B Instruct 262K 2bit256K / 2.5 GB111
...B Instruct 262k V2 EXL2 6.0bpw256K / 6.7 GB41
Note: green Score (e.g. "73.2") means that the model is better than bullerwins/gradientai_Llama-3-8B-Instruct-262k_v2_exl2_5.0bpw.