← back to catalog · registered 2026-11-15 19:03

anaaaaaa13421312/gemma-4-31B-it-abliterated-Japanese-RP-FP8-Dynamic

anaaaaaa13421312 Gemma 31B 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/anaaaaaa13421312%2Fgemma-4-31B-it-abliterated-Japanese-RP-FP8-Dynamic"
Response includes
  • classification m1
  • files 12
  • author_summary 4 models
  • readme_text full
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)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
Downloads · 30-day
84
Likes
0
Model age
6d ago
created 2026-11-15
Downloads over time
Now 84 → from 0 ↑ 0%
0316292 0 on Nov 15 84 on Nov 22 Nov
Nov 15 → Nov 22 · 8 snapshots · spans 7 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.

Metadata

License
apache-2.0
Tags
llm-compressor safetensors gemma4 compressed-tensors fp8_dynamic japanese roleplay not-for-all-audiences abliterated conversational text-generation ja

Related

Total size
31.0 GB
Files
12
Quantizations
1
Registered
2026-11-15 19:03
Last updated on HF
2026-11-17 08:39

Files by quantization

Auxiliary files 12 files 31.0 GB
model-00002-of-00002.safetensors 15.5 GB 780bb0aa download
model-00001-of-00002.safetensors 15.4 GB 68022369 download
tokenizer.json 30.7 MB cc8d3a0c download
model.safetensors.index.json 158 KB c14d05b3 download
chat_template.jinja 16.5 KB f62ca843 download
DEPRECATED–INCOMPLETE–CONTAINS_ERRORS_ERRORS.README.md 11.9 KB 705579de download
LICENSE 11.1 KB d6456956 download
tokenizer_config.json 6.83 KB 38ca505f download
config.json 6.37 KB 080a6924 download
.gitattributes 1.53 KB 52373fe2 download
README.md 818 B 6eec4518 download
generation_config.json 204 B 72c589a8 download

README current version from Hugging Face


license: apache-2.0
base_model: anaaaaaa13421312/gemma-4-31B-it-abliterated-Japanese-RP
base_model_relation: quantized
library_name: llm-compressor
tags:

  • compressed-tensors
  • gemma4
  • fp8_dynamic
  • japanese
  • roleplay
  • not-for-all-audiences

gemma-4-31B-it-abliterated-Japanese-RP-FP8-Dynamic

anaaaaaa13421312/gemma-4-31B-it-abliterated-Japanese-RP を llmcompressor.model_free_ptq(scheme="FP8_DYNAMIC")でデータフリー量子化したチェックポイントです。

  • Format: compressed-tensors
  • Scheme: FP8_DYNAMIC
  • Target: FP8 (E4M3 per-channel static weights + dynamic per-token activations) for RunPod 48GB GPUs (L40S / RTX 6000 Ada / A40 Marlin) and TPU v7x.
  • Unquantized (BF16 preserved): lm_head, vision/audio modules, embed_tokens, norms

README history 4 versions

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

  1. 2026-11-17 Embed normalized learning curve plots in bilingual model cards aca5d56 60 KB
    Loading...
  2. 2026-11-16 Update bilingual model cards (README.md, README_ja.md) 7a06ad6 59.9 KB
    Loading...
  3. 2026-11-15 Update README.md dad4e01 818 B
    Loading...
  4. 2026-11-15 Upload FP8_DYNAMIC quantized checkpoint (2 shards) 6e5513a 740 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.

Open in Abliteration