← back to catalog · registered 2026-11-11 17:03

onemunanoyo/LFM3-VL-2B-Abliterated-GGUF

onemunanoyo 2B GGUF multimodal
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/onemunanoyo%2FLFM3-VL-2B-Abliterated-GGUF"
Response includes
  • classification m8
  • files 10
  • author_summary 8 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M8
Primary method

Repackaging (quantization)

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.
  • 'abliterated' in name/tags
  • is_gguf=1
  • assume M1 (base ablation) + M8 (GGUF quant) - default when producer unknown
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 · 30-day
480
Likes
0
Model age
1w ago
created 2026-11-10
Downloads over time
Now 480 → from 0 ↑ 0%
0176352528 0 on Nov 11 480 on Nov 22 Nov
Nov 11 → Nov 22 · 12 snapshots · spans 11 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
other
Quantizations
BF16 Q4_K Q6_K Q8_0
Tags
gguf Abliterated image-text-to-text base_model:LiquidAI/LFM3-VL-2B base_model:quantized:LiquidAI/LFM3-VL-2B license:other endpoints_compatible region:us conversational

Related

Total size
7.30 GB
Files
10
Quantizations
5
Registered
2026-11-11 17:03
Last updated on HF
2026-11-11 16:31

Files by quantization

BF16 2 files 3.58 GB
LFM3-2B-VL-Abliterated-BF16.gguf 3.23 GB a0813772 download
mmproj-LFM3-2B-VL-Abliterated-BF16.gguf 361 MB f85c3de0 download
Q8_0 1 file 1.72 GB
LFM3-2B-VL-Abliterated-Q8_0.gguf 1.72 GB e97309bb download
Q6_K 1 file 1.33 GB
LFM3-2B-VL-Abliterated-Q6_K.gguf 1.33 GB 4e1abe2b download
Q4_K 1 file 1.02 GB
LFM3-2B-VL-Abliterated-Q4_K_M.gguf 1.02 GB a8c44559 download
Auxiliary files 5 files 17.1 KB
LICENSE 10.3 KB 272d2349 download
README.md 4.10 KB 45b08ba9 download
.gitattributes 1.83 KB 497d4ec1 download
sha256.txt 503 B a7c2bd3a download
NOTICE-MODIFICATIONS.md 373 B afea2619 download

README current version from Hugging Face


license_name: lfm1.0
license_link: LICENSE
license: other
base_model:

  • LiquidAI/LFM3-VL-2B
    tags:
  • Abliterated
  • image-text-to-text

LFM3-VL-2B-Abliterated-GGUF

GGUF builds of an abliterated derivative of LiquidAI/LFM3-VL-2B

Benchmarks

All numbers below were measured on a user-provided Japanese evaluation set (84 harmful/NSFW prompts and 8 benign prompts).
Refusal labels come from a separate 3B decision model (d1-3B), not from string matching.

Sampling: temperature 0.1, top_k 50, top_p 1.0, repetition_penalty 1.1, thinking ON, max_new 4096.

Refusal (84 prompts, thinking ON)

Model Refusals
LFM3-VL-2B (base) 58 / 84
This model (BF16) 1 / 84

Quantization (llama.cpp llama-perplexity, shared JA/EN corpus, 512 ctx)

File PPL vs. BF16 Refusals (84 prompts)
BF16 38.78 — 1 / 84
Q8_0 38.74 −0.1% 0 / 84
Q6_K 38.50 −0.7% 0 / 84
Q4_K_M 42.17 +8.7% 2 / 84

Files

File Size Notes
LFM3-2B-VL-Abliterated-BF16.gguf 3.47 GB Original weights
LFM3-2B-VL-Abliterated-Q8_0.gguf 1.85 GB Recommended (practically lossless)
LFM3-2B-VL-Abliterated-Q6_K.gguf 1.43 GB Smaller, PPL slightly below BF16
LFM3-2B-VL-Abliterated-Q4_K_M.gguf 1.10 GB Smallest, +8.7% PPL and 2 refusals
mmproj-LFM3-2B-VL-Abliterated-BF16.gguf 379 MB Vision projector — required for image input (pairs with every quant above)

Run (llama.cpp)

Requires llama.cpp b12040 or newer. Older builds fail with unknown model architecture: 'lfm3'.

llama-server -m LFM3-2B-VL-Abliterated-Q8_0.gguf \
             --mmproj mmproj-LFM3-2B-VL-Abliterated-BF16.gguf \
             --jinja --reasoning-format none -ngl 99 -c 8192

Notes:

  • Without --mmproj the model runs text-only; image inputs are rejected.
  • Thinking is ON by default (the chat template starts the assistant turn with thinking). Disable it with "chat_template_kwargs": {"enable_thinking": false}.
  • Use max_tokens >= 4096. Thinking is long (median ~3.4k tokens on the eval set); with 2048 the answer is often cut off mid-thought.
  • Sampling used for the benchmarks: temperature 0.1, top_k 50, top_p 1.0, repetition_penalty 1.1.
  • --reasoning-format none keeps the thinking inside message.content, matching the transformers harness. The default splits it into reasoning_content.

Changes

  • Weight edit only. No training, fine-tuning, or distillation.
  • Removed a refusal direction (strength α = 1.3) from 40 residual-writer matrices in layers 4–23: self_attn.o_proj (10), conv.out_proj (10), and feed_forward.w2 (20).
  • Direction: per-layer mean difference of hidden states between 416 harmful and 416 harmless English prompts, taken at the last prompt token, thinking enabled. The projection is computed in FP32, and the weights are saved in BF16.
  • Orthogonalization is sandwich-norm aware: this model adds each writer's output to the residual after post_operator_norm / post_ffn_norm. The projection therefore targets g ⊙ u (g = the norm's learned scale) instead of u, so the removed component does not return after the norm.
  • Untouched: embed_tokens, all norms, q/k/v_proj, conv.in_proj, conv.conv, attn.gate_proj, w1/w3, lm_head (tied) — and the whole vision tower / multimodal projector, which are byte-identical to upstream (384 of 424 tensors unchanged). The mmproj is therefore the upstream vision encoder, unchanged by this derivative.
  • Relative weight change: mean 2.97e-04, max 5.37e-04; max |ΔW| = 5.49e-01.
  • Raw-text perplexity of this model is +1.9% vs. base (95% CI −1.5 to +5.3%). Instruction-copy perplexity looks better than base, but that metric saturates and is not a quality gate.

License

Released under the LFM Open License v1.0 (copy included as LICENSE). This is a derivative work of LiquidAI/LFM3-VL-2B; the files were modified, as stated in NOTICE-MODIFICATIONS.

  • Commercial use is limited to entities with annual revenue below 10 million USD.
  • Liquid AI does not endorse or support this model.

README history 6 versions

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

  1. 2026-11-11 Update README.md 162de9e 4.1 KB
    Loading...
  2. 2026-11-11 Update README.md b85f352 4.1 KB
    Loading...
  3. 2026-11-11 Update README.md ac0b662 4.1 KB
    Loading...
  4. 2026-11-11 Update README.md 1a0222f 4.1 KB
    Loading...
  5. 2026-11-11 Upload 7 files c6413b6 4.7 KB
    Loading...
  6. 2026-11-10 initial commit 1f66414 66 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