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

Llama 3 8B Instruct Abliterated by failspy

 »  All LLMs  »  failspy  »  Llama 3 8B Instruct Abliterated   URL Share it on

  7   
  Autotrain compatible   Conversational   Endpoints compatible   Instruct   Llama   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/failspy/Llama-3-8B-Instruct-abliterated 

Llama 3 8B Instruct Abliterated Benchmarks

MMLU Pro: 19.35
GPQA: 3.47
MUSR: 10.51
BBH: 18.86
IFEval: 59.09 vs 88 (so35)-32.9%
ARC: 51.45 vs 96.7 (so35)-46.8%
HellaSwag: 73.97 vs 95.3 (gpt4)-22.4%
MMLU: 56.16 vs 88.3 (so35)-36.4%
TruthfulQA: 47.9 vs 59 (gpt4)-18.8%
WinoGrande: 71.43 vs 87.5 (gpt4)-18.4%
GSM8K: 41.47 vs 96.4 (so35)-57%
MATH Lvl 5: 3.85
LLME Score: 0.23074
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Llama 3 8B Instruct Abliterated (failspy/Llama-3-8B-Instruct-abliterated)
🌟 Advertise your project 🚀

Llama 3 8B Instruct Abliterated Parameters and Internals

Model Type 
text generation
Additional Notes 
This model may come with interesting quirks due to the orthogonalization applied to inhibit refusal. It is important to recognize that while it aims to reduce refusal, it is not guaranteed to never refuse or to consistently understand requests. It retains all other tuning aspects of the original model. Further testing and exploration of its quirks and potential side effects is encouraged.
LLM NameLlama 3 8B Instruct Abliterated
Repository 🤗https://huggingface.co/failspy/Llama-3-8B-Instruct-abliterated 
Model Size8b
Required VRAM16.1 GB
Updated2025-09-23
Maintainerfailspy
Model Typellama
Instruction-BasedYes
Model Files  5.0 GB: 1-of-4   5.0 GB: 2-of-4   4.9 GB: 3-of-4   1.2 GB: 4-of-4
Model ArchitectureLlamaForCausalLM
Licensellama3
Context Length8192
Model Max Length8192
Transformers Version4.41.0.dev0
Tokenizer ClassPreTrainedTokenizerFast
Vocabulary Size128256
Torch Data Typebfloat16

Best Alternatives to Llama 3 8B Instruct Abliterated

Best Alternatives
Context / RAM
Downloads
Likes
...otron 8B UltraLong 4M Instruct4192K / 32.1 GB2522120
UltraLong Thinking4192K / 16.1 GB1153
...a 3.1 8B UltraLong 4M Instruct4192K / 32.1 GB17624
...otron 8B UltraLong 2M Instruct2096K / 32.1 GB114215
...a 3.1 8B UltraLong 2M Instruct2096K / 32.1 GB8759
...otron 8B UltraLong 1M Instruct1048K / 32.1 GB714252
...a 3.1 8B UltraLong 1M Instruct1048K / 32.1 GB138729
Zero Llama 3.1 8B Beta61048K / 16.1 GB21
...dger Nu Llama 3.1 8B UltraLong1048K / 16.2 GB53
....1 1million Ctx Dark Planet 8B1048K / 32.3 GB83
Note: green Score (e.g. "73.2") means that the model is better than failspy/Llama-3-8B-Instruct-abliterated.

Rank the Llama 3 8B Instruct Abliterated 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? 51535 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 v20241124
  Support LLM Explorer