CUSTOM
CLANKERS.
← Browse all models

AtomicChat / Model details

LFM2.5-24B-A2B-Uncensored-GGUF

AtomicChat/LFM2.5-24B-A2B-Uncensored-GGUF

uncensoredfine-tuned
Download files on Hugging Face ↗

Read the original model card ↗ · Choosing a model

About this model

AtomicChat's refusal-modified derivative of Liquid AI's LFM2.5-24B-A2B, shipped as GGUF weights. The creator describes projecting a refusal direction out of 52 matrices in blocks 14-39 at scale 1.75, leaving the earlier blocks, router and embeddings untouched, and ships six baked GGUF builds (12.1 to 25.7 GB) plus a 24 MB rank-1 LoRA adapter meant for existing quants of the original weights. According to the model card, held-out testing showed English refusals falling from 96.3% to 1.2% with MMLU essentially unchanged (64.40% to 65.55%), and the creator recommends the 15.4 GB AD-Q4_K_M build for 24 GB machines; these figures are the creator's own and have not been independently verified here. The card cautions that answers are phrased differently from the original, that only English and Russian refusals were measured, and that removing refusals does not make output safe, correct or lawful. The metadata lists the license as other, and the card names the LFM Open License v1.0.

Description sources

Source-based description, not an independent evaluation. Check the linked sources for the creator’s claims and limitations.

Size and formats

Base model
LiquidAI/LFM2.5-24B-A2B
Parameters
Unknown
Main repository formats
GGUF
License tag
other
Languages
ar, de, en, es, fr, it, ja, ko, pt, zh
Last updated
2026-10-13 · 20:02:10.000 UTC

Community reviews & tips

Source-linked paraphrases of sampled community sources, not independently tested by this directory.

Sample window: to · 1 unique source link. This is a selected sample, not all community discussions.

Small sample: source links are not a count of independent reviewers or evidence of consensus.

Sources checked:

What changed

Model types are creator-reported, not independently tested. These labels describe source metadata, not a guarantee of behavior.

  • uncensoredmodel name: LFM2.5-24B-A2B-Uncensored-GGUF
  • fine-tunedsource tag: base_model:finetune:LiquidAI/LFM2.5-24B-A2B

Family evidence

These groupings are source-reported. Explicit base identifiers take priority over generic architecture tags.

  • LFM: Base model: LiquidAI/LFM2.5-24B-A2B

Available versions

Versions appear here only when they explicitly name this exact model as their source. Check each repository for available files.

Refusal profile

Refused 0 of 18 prompts (0.0%)

We asked this model 120 benign but risky-looking prompts across ten topics and counted how often it refused, hedged or answered. Lower means it answered more often; it says nothing about answer quality.

  • Persuasion & deception0 refused · 0 partial · 2 complied
  • Harassment & confrontation0 refused · 0 partial · 2 complied
  • Risky activities0 refused · 0 partial · 2 complied
  • Offensive & controversial speech0 refused · 0 partial · 1 complied
  • Legal grey areas0 refused · 0 partial · 1 complied
  • Privacy & personal data0 refused · 0 partial · 1 complied
  • Self-harm & mental health topics0 refused · 0 partial · 2 complied
  • Sexual & adult topics0 refused · 0 partial · 2 complied
  • Ethically grey requests0 refused · 0 partial · 2 complied
  • Violence & conflict0 refused · 0 partial · 3 complied

Capability check: 13 of 15 passed. Fifteen simple tasks with known answers, so a broken model cannot score well by refusing nothing.

Unclear answers: 102 (not counted). An answer is unclear when the two checker models disagree or the output was empty or garbled.

Too many unclear answers to rank: this model appears on the Refusal Index but is not placed in the ranking.

Measured 2026-10-25 · Q4_K_M via AtomicChat/LFM2.5-24B-A2B-Uncensored-GGUF ↗ · llama.cpp · prompt set v1

How we measure · The prompts

Link to this listing

Made or quantized this model? Paste one of these into your model card or README so people can find this listing.

Markdown

[Listed on Custom Clankers](https://customclankers.com/models/atomicchat-lfm2-5-24b-a2b-uncensored-gguf/)

HTML

<a href="https://customclankers.com/models/atomicchat-lfm2-5-24b-a2b-uncensored-gguf/">Listed on Custom Clankers</a>

Plain text links only: no image, no script, nothing loaded from this site.

Customize or build on this model

Listed from public GitHub data, not tested by us. Read the code and license before you run anything.

Tools to customize it further

Make a smaller download (quantize)

  • llama.cpp ↗ ★ 132,260

    Runs models locally and includes the tools behind most GGUF files. A conversion script turns Hugging Face weights into GGUF, and llama-quantize shrinks a high-precision GGUF to lower-bit formats, which can cost some accuracy.

Learn the techniques

  • LLM Course ↗ ★ 84,667

    A free course with roadmaps and Colab notebooks. Its notebooks include one-click tools for merging with MergeKit, quantizing to GGUF, GPTQ, EXL2, AWQ and HQQ, and automatic abliteration with custom datasets.

Compare all tools →

Source provenance

Downloads
51,478
Likes
66
Metadata checked
2026-10-24 · 09:40:11 UTC
First seen in index
2026-10-24 · 04:11:48 UTC
Source revision
1acb8f6eb027d85b7771dcc9eb68817aa480e9c8

Source-card excerpt

Source model card excerpt: How to Run LFM2.5-24B-A2B Without Refusals Locally Liquid's refusal behaviour projected out of the original weights, in the second two thirds of the network

From AtomicChat/LFM2.5-24B-A2B-Uncensored-GGUF. Creator model-card excerpt, not an independent review.

Inspect before you run

Low refusal is not proof of correctness. Outputs can be inaccurate, harmful, or biased. Review the model card, license and files yourself. A license tag is not commercial-use advice. Parameter counts are source metadata; no hardware compatibility is implied.

This directory does not host or redistribute model weights. Refusal profiles are measured by us on a fixed prompt set; all other behavior claims are source-reported.

Collection rules & limitations ↗