← back to catalog · registered 2026-11-20 10:04

aug11/DeepSeek-V4-Flash-0731-MXFP4-MLX-abliterated

aug11 Deepseek
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/aug11%2FDeepSeek-V4-Flash-0731-MXFP4-MLX-abliterated"
Response includes
  • classification m1
  • files 44
  • author_summary 2 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
44
Likes
0
Model age
2d ago
created 2026-11-20
Downloads over time
Now 44 → from 0 ↑ 0%
0163248 0 on Nov 20 44 on Nov 22 Nov
Nov 20 → Nov 22 · 3 snapshots · spans 2 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
mit
Tags
mlx safetensors deepseek_v4 abliterated mxfp4 text-generation base_model:TensorFold/DeepSeek-V4-Flash-0731-MXFP4-MLX base_model:quantized:TensorFold/DeepSeek-V4-Flash-0731-MXFP4-MLX license:mit 4-bit region:us

Related

Total size
156 GB
Files
44
Quantizations
1
Registered
2026-11-20 10:04
Last updated on HF
2026-11-20 09:38

Files by quantization

Auxiliary files 44 files 156 GB
model-00021.safetensors 4.66 GB 03f6a35c download
model-00002.safetensors 4.66 GB efbe1300 download
model-00032.safetensors 4.66 GB 04c8bce8 download
model-00033.safetensors 4.66 GB 1c60bc61 download
model-00017.safetensors 4.66 GB dac3eae9 download
model-00010.safetensors 4.66 GB faeb37e9 download
model-00030.safetensors 4.66 GB 927fd42c download
model-00020.safetensors 4.66 GB a8b68425 download
model-00023.safetensors 4.66 GB d173767e download
model-00013.safetensors 4.66 GB 0bea3b4d download
model-00007.safetensors 4.66 GB f43dd971 download
model-00015.safetensors 4.66 GB 8a3f9a31 download
model-00012.safetensors 4.66 GB e585985b download
model-00025.safetensors 4.66 GB c3253614 download
model-00028.safetensors 4.66 GB c67e8019 download
model-00018.safetensors 4.66 GB 6ba9b7b8 download
model-00031.safetensors 4.66 GB f54f81ab download
model-00008.safetensors 4.66 GB f87e2ab2 download
model-00005.safetensors 4.66 GB 88c3c3b7 download
model-00001.safetensors 4.66 GB 4636aa0e download
model-00026.safetensors 4.66 GB 0403b870 download
model-00027.safetensors 4.66 GB 50df9348 download
model-00022.safetensors 4.66 GB b9ce3556 download
model-00009.safetensors 4.66 GB 53aef381 download
model-00014.safetensors 4.66 GB e7227fbc download
model-00019.safetensors 4.66 GB ec5a01fd download
model-00029.safetensors 4.66 GB e6ab71e7 download
model-00024.safetensors 4.66 GB 6e49cf80 download
model-00016.safetensors 4.66 GB ca28080b download
model-00011.safetensors 4.66 GB be6755de download
model-00004.safetensors 4.66 GB 571ea55b download
model-00006.safetensors 4.66 GB f7a6af34 download
model-00003.safetensors 4.66 GB f56932d1 download
model-00034.safetensors 1.86 GB 0c588dbe download
tokenizer.json 6.07 MB 628e3364 download
model.safetensors.index.json 4.76 MB 9530a6a3 download
config.json 102 KB 56d04fc6 download
deepseek_v4.py 52.4 KB accc7a77 download
README.md 2.98 KB 4697d483 download
generate.py 2.80 KB 2d598443 download
.gitattributes 1.48 KB a6344aac download
LICENSE 1.06 KB d62e3bef download
tokenizer_config.json 801 B f3dad388 download
generation_config.json 170 B c56a8c5b download

README current version from Hugging Face


license: mit
base_model:

  • TensorFold/DeepSeek-V4-Flash-0731-MXFP4-MLX
  • prem-research/DeepSeek-V4-Flash-0731-abliterated
    library_name: mlx
    tags:
  • mlx
  • abliterated
  • deepseek_v4
  • mxfp4
    pipeline_tag: text-generation

DeepSeek-V4-Flash-0731 MXFP4-MLX — abliterated

This is prem-research/DeepSeek-V4-Flash-0731-abliterated
(revision 457210de) in the MLX layout of
TensorFold/DeepSeek-V4-Flash-0731-MXFP4-MLX (revision d6be051a).
Both derive from deepseek-ai/DeepSeek-V4-Flash-0731.

How it was made

TensorFold/DeepSeek-V4-Flash-0731-MXFP4-MLX is a byte transplant of deepseek-ai/DeepSeek-V4-Flash-0731: fp4 e2m1 expert weights stored as MLX
mxfp4 with the same bytes, fp8 e4m3 dense weights as MLX mxfp8 with the same bytes and each 128x128
block scale repeated per 32-group, every other tensor unchanged under the same name.
prem-research/DeepSeek-V4-Flash-0731-abliterated keeps the original format and rewrites only the shards of layers 17-36.
Method: ARA (prem-research model card): attn.wo_b, ffn.shared_experts.w2 and the listed experts' ffn.experts.E.w2, layers 17-36.

This repository is the base with the tensors that changed replaced by the same transplant:

  • every tensor of the rewritten source shards was transplanted and compared with the base:
    320 of 31,450 differ, all inside the steered modules; the other
    31,130 are byte-identical to the base, which re-proves the transplant on this data;
  • the 320 changed tensors were written in place into 15 base shards;
    headers, offsets, shapes and dtypes did not move;
  • every other tensor and file is byte-identical to TensorFold/DeepSeek-V4-Flash-0731-MXFP4-MLX at d6be051a,
    except this README;
  • numeric check: each changed module dequantized with mx.dequantize from this layout against the
    fp8/fp4 source, max abs error 0.0.

Recipe: byte transplant: fp4 I8->U32 and fp8 F8_E4M3->U32 same bytes; fp4 scales same bytes; fp8 128x128 block scales repeated to [R, C/32]; other tensors unchanged.
The splice is scripts/lab/dsv4/dsv4_abliterated_splice.py, run by
scripts/lab/ablit-perf/job-dsv4-abliterated.sh in the MacProvider repository.

Checks this artifact passed

The publish step runs only after the job passed all of its gates:

  • token parity of the MacProvider native engine against DeepSeek's reference implementation on these
    weights at 1K and 8K context: first 16 top-1 tokens exact and at least 240 of 256 greedy tokens equal;
  • refusal sanity (greedy, keyword classes): the base refused 20 of 20 harmful prompts;
    this model refused 0 of 20 harmful and 0 of 20 harmless prompts.

These are lab sanity gates, not a capability or safety evaluation.

License

MIT, the license of both TensorFold/DeepSeek-V4-Flash-0731-MXFP4-MLX and prem-research/DeepSeek-V4-Flash-0731-abliterated (and of deepseek-ai/DeepSeek-V4-Flash-0731).

README history 1 version

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

  1. 2026-11-20 Abliterated DeepSeek-V4-Flash-0731 MXFP4-MLX 93580c7 3 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