Meta Llama 3 8B Instruct Abliterated V3 By failspy: Benchmarks, Features and Detailed Analysis. Insights on Meta Llama 3 8B Instruct Abliterated V3.

Autotrain compatible Conversational Endpoints compatible Instruct Llama Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/failspy/Meta-Llama-3-8B-Instruct-abliterated-v3

Meta Llama 3 8B Instruct Abliterated V3 Benchmarks

ARC: 61.6 vs 96.7 (so35)^-36.3%

HellaSwag: 78.35 vs 95.3 (gpt4)^-17.8%

MMLU: 66.67 vs 88.3 (so35)^-24.5%

TruthfulQA: 52.44 vs 59 (gpt4)^-11.1%

WinoGrande: 74.51 vs 87.5 (gpt4)^-14.8%

GSM8K: 70.05 vs 96.4 (so35)^-27.3%

LLME Score: 0.2816

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

Meta Llama 3 8B Instruct Abliterated V3 (failspy/Meta-Llama-3-8B-Instruct-abliterated-v3)

Meta Llama 3 8B Instruct Abliterated V3 Parameters and Internals

Model Type

text generation

Additional Notes

The model has undergone orthogonalization to remove refusal behavior without altering other aspects of the original 70B instruct model. Known as "abliteration," this process differentiates it from standard uncensored fine-tunes. Potential quirks due to new methodology are acknowledged.

Training Details

Methodology:

Orthogonalization (Ablation) of refusal behavior based on a refined methodology.

LLM Name	Meta Llama 3 8B Instruct Abliterated V3
Repository 🤗	https://huggingface.co/failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
Model Size	8b
Required VRAM	16.1 GB
Updated	2024-12-22
Maintainer	failspy
Model Type	llama
Instruction-Based	Yes
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 Architecture	LlamaForCausalLM
License	llama3
Context Length	8192
Model Max Length	8192
Transformers Version	4.40.1
Tokenizer Class	PreTrainedTokenizerFast
Vocabulary Size	128256
Torch Data Type	bfloat16

Quantized Models of the Meta Llama 3 8B Instruct Abliterated V3

Model	Likes	Downloads	VRAM
...truct Abliterated V3 GPTQ 4bit	0	9	5 GB

Best Alternatives to Meta Llama 3 8B Instruct Abliterated V3

Best Alternatives	Context / RAM	Downloads	Likes
...a 3 8B Instruct Gradient 1048K	1024K / 16.1 GB	4850	678
MrRoboto ProLong 8B V4b	1024K / 16.1 GB	94	0
...o ProLongBASE Pt6 Unaligned 8B	1024K / 16.1 GB	60	0
MrRoboto ProLong 8B V4c	1024K / 16.1 GB	79	0
MrRoboto ProLong 8B V1a	1024K / 16.1 GB	107	0
MrRoboto ProLong 8B V2a	1024K / 16.1 GB	101	0
...o ProLongBASE Pt2 Unaligned 8B	1024K / 16.1 GB	45	0
MrRoboto ProLong 8B V2f	1024K / 16.1 GB	74	0
MrRoboto ProLong 8B V4f	1024K / 16.1 GB	45	0
MrRoboto ProLong 8B V1l	1024K / 16.1 GB	67	0

Note: green Score (e.g. "73.2") means that the model is better than failspy/Meta-Llama-3-8B-Instruct-abliterated-v3.

Rank the Meta Llama 3 8B Instruct Abliterated V3 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? 40123 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