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

djuna/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2-remap

djuna Qwen 15B 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/djuna%2FDeepSeek-R1-Distill-Qwen-14B-abliterated-v2-remap"
Response includes
  • classification m5
  • files 17
  • benchmarks 5 entries
  • hub_downloads_all_time 173
  • author_summary 2 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M5
Primary method

Mergekit merge

Applied on top of direct removal inherited from the base model.
Confidence
MEDIUM
Inherited from base model
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.
  • merge tag / mergekit / dare-ties in tags or name
  • no unusual architecture pattern (regular merge)
  • abliterated marker present
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · lifetime
173
38 last 30d - stable
Likes
2
Descendants
3
in 3 direct forks
Model age
19mo ago
created 2025-03-11
Downloads over time
Now192→from0↑0%
0701412110 on Mar 5, 2025192 on Oct 11Mar '25Jun '25Sep '25Dec '25MarJunSep
Mar 5, 2025 → Oct 11 · 123 snapshots · spans 585 days

Benchmarks

Benchmark Score Source
BBH average 0.3473861720067454 OpenLLM-v2
IFEval instruct 0.4892086330935252 OpenLLM-v2
IFEval-Prompt 0.35304990757855825 OpenLLM-v2
MATH lvl 5 0 OpenLLM-v2
MMLU-Pro 0.19148936170212766 OpenLLM-v2

Genealogy 3 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

Tags
transformers safetensors qwen2 text-generation mergekit merge conversational arxiv:2203.05482 base_model:djuna/Q2.5-Veltha-14B-0.5 base_model:merge:djuna/Q2.5-Veltha-14B-0.5 base_model:huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2 base_model:merge:huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2

Related

Total size
27.5 GB
Files
17
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2025-03-11 23:11

Files by quantization

Auxiliary files 17 files 27.5 GB
model-00002-of-00006.safetensors 4.61 GB 71c60429 download
model-00003-of-00006.safetensors 4.61 GB 399ea518 download
model-00004-of-00006.safetensors 4.61 GB 116bd9c7 download
model-00005-of-00006.safetensors 4.61 GB cd9fdcc0 download
model-00001-of-00006.safetensors 4.56 GB af403ad4 download
model-00006-of-00006.safetensors 4.48 GB 7cd23d22 download
tokenizer.json 10.9 MB 8da66dbb download
vocab.json 2.65 MB 4783fe10 download
merges.txt 1.59 MB 31349551 download
model.safetensors.index.json 44.1 KB 74c0cf96 download
tokenizer_config.json 8.05 KB 96e46958 download
README.md 2.84 KB 114b9159 download
mergekit_config.yml 2.06 KB 7f1831ca download
.gitattributes 1.53 KB 52373fe2 download
added_tokens.json 810 B fb9d69f3 download
config.json 764 B fd3aad03 download
special_tokens_map.json 613 B ac23c0aa download

README current version from Hugging Face


base_model:

  • huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2
  • djuna/Q2.5-Veltha-14B-0.5
    library_name: transformers
    tags:
  • mergekit
  • merge

merge

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the Linear merge method using huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2 as a base.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2
    parameters:
      weight: 1
  - model: djuna/Q2.5-Veltha-14B-0.5
    parameters:
      weight: 0
merge_method: linear
base_model: huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2
parameters:
  normalize: false
tokenizer:
  source: "djuna/Q2.5-Veltha-14B-0.5"
  tokens: 
    <|endoftext|>:
      source: "djuna/Q2.5-Veltha-14B-0.5"
    <|im_start|>:
      source: "djuna/Q2.5-Veltha-14B-0.5"
    <|im_end|>:
      source: "djuna/Q2.5-Veltha-14B-0.5"
    <|object_ref_start|>:
      source: "djuna/Q2.5-Veltha-14B-0.5"
    <|object_ref_end|>:
      source: "djuna/Q2.5-Veltha-14B-0.5"
    <|box_start|>:
      source: "djuna/Q2.5-Veltha-14B-0.5"
    <|box_end|>:
      source: "djuna/Q2.5-Veltha-14B-0.5"
    <|end▁of▁sentence|>:
      source: 
        model: "huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2"
        kind: "model_token"
        token: "<|end▁of▁sentence|>"
      force: true
    <|User|>:
      source: 
        model: "huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2"
        kind: "model_token"
        token: "<|User|>"
      force: true
    <|Assistant|>:
      source: 
        model: "huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2"
        kind: "model_token"
        token: "<|Assistant|>"
      force: true
    <|begin▁of▁sentence|>:
      source: 
        model: "huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2"
        kind: "model_token"
        token: "<|begin▁of▁sentence|>"
      force: true
    <|EOT|>:
      source: 
        model: "huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2"
        kind: "model_token"
        token: "<|EOT|>"
      force: true
    <think>:
      source: 
        model: "huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2"
        kind: "model_token"
        token: "<think>"
      force: true
    </think>:
      source: 
        model: "huihui-ai/DeepSeek-R1-Distill-Qwen-14B-abliterated-v2"
        kind: "model_token"
        token: "</think>"
      force: true
dtype: bfloat16

README history 1 version

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

  1. 2025-03-11Upload folder using huggingface_hub967d9942.8 KB
    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