Get the app
← back to catalog · registered 2026-08-22 13:56

huggingkot/Llama-3-8B-Instruct-abliterated-v2-bnb-4bit

huggingkot Llama 7.2B second-order
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model. 

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/huggingkot%2FLlama-3-8B-Instruct-abliterated-v2-bnb-4bit"
Response includes
  • classification m1
  • files 10
  • hub_downloads_all_time 882
  • author_summary 8 models
  • readme_text full
Try next
GET/models?author=huggingkot
All 8 models by huggingkot
 GET/models?family=llama&method=m1
Sibling llama models · same method
 GET/methods/m1
What "m1" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
How we label → About M1 →
Refusal direction extracted via
Extraction technique

Difference-of-means

FromArditi et al. (2024) ↗
Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
What is a refusal direction? → Read the paper ↗
Downloads · lifetime
882
327 last 30d - stable
Likes
0
Model age
19mo ago
created 2025-03-06
Downloads over time
Now1.1K→from54↑1,928%
24018001.2K54 on Mar 5, 20251.1K on Oct 11Mar '25Jun '25Sep '25Dec '25MarJunSep
Mar 5, 2025 → Oct 11 · 123 snapshots · spans 585 days

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

cognitivecomputations/Llama-3-8B-Instruct-abliterated-v2 this lineage
huggingkot/Llama-3-8B-Instruct-abliterated-v2-bnb-4bit↓ 327

Metadata

Tags
safetensors base_model:QuixiAI/Llama-3-8B-Instruct-abliterated-v2 base_model:finetune:QuixiAI/Llama-3-8B-Instruct-abliterated-v2 8-bit region:us

Related

Other by huggingkot
Llama-3-8B-Instruct-abliterated-v2-Q4_K_M-GGUF 253 NeuralDaredevil-8B-abliterated-q4f16_1-MLC 96 dolphin-2.9.2-Phi-3-Medium-abliterated-Q4_K_M-GGUF 58 Llama-3-8B-Instruct-abliterated-v2-q4f16_1-MLC 23 NeuralDaredevil-8B-abliterated-bnb-4bit 15
Same base model
featherless-ai-quants/cognitivecomputations-Llama-3-8B-Instruct-abliterated-v2-GGUF 560 tensorblock/Llama-3-8B-Instruct-abliterated-v2-GGUF 314 huggingkot/Llama-3-8B-Instruct-abliterated-v2-Q4_K_M-GGUF 253 solidrust/Llama-3-8B-Instruct-abliterated-v2-AWQ 23 huggingkot/Llama-3-8B-Instruct-abliterated-v2-q4f16_1-MLC 23
Total size
5.31 GB
Files
10
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2025-03-06 03:51
Source huggingface.co/huggingkot/Llama-3-8B-Instruct-abliterated-v2-bnb-4bit ↗ Base model cognitivecomputations/Llama-3-8B-Instruct-abliterated-v2 ↗

Files by quantization

Auxiliary files 10 files 5.33 GB
model-00001-of-00002.safetensors 4.33 GB 3232a8f7 download
model-00002-of-00002.safetensors 1002 MB 8005050c download
tokenizer.json 16.4 MB 3c5cf440 download
model.safetensors.index.json 129 KB 49ecc467 download
tokenizer_config.json 49.9 KB 2f72bef9 download
.gitattributes 1.53 KB c4cb545c download
config.json 1.18 KB 690728b3 download
README.md 920 B ddbf6972 download
special_tokens_map.json 345 B 6cc5fa1d download
generation_config.json 220 B c72858f1 download

README current version from Hugging Face

base_model:

  • cognitivecomputations/Llama-3-8B-Instruct-abliterated-v2

This is a converted weight from Llama-3-8B-Instruct-abliterated-v2 model in unsloth 4-bit dynamic quant using this collab notebook.

About this Conversion

This conversion uses Unsloth to load the model in 4-bit format and force-save it in the same 4-bit format.

How 4-bit Quantization Works

  • The actual 4-bit quantization is handled by BitsAndBytes (bnb), which works under Torch via AutoGPTQ or BitsAndBytes.
  • Unsloth acts as a wrapper, simplifying and optimizing the process for better efficiency.

This allows for reduced memory usage and faster inference while keeping the model compact.

README history 1 version

The author's README evolved over time. Click a version to see its content at that point.

  1. 2025-03-06add files247b3b9920 B
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration