🤖 LFM2.5-VL-450M-heretic — Importance Matrix GGUF
This repository hosts importance-matrix (imatrix) optimized GGUF weights, available in multiple quantization formats, and the associated vision projection matrix for
LFM2.5-VL-450M-heretic, quantized from the source floating-point tensors provided by
coder3101/LFM2.5-VL-450M-heretic.
🔄 Sister Repository: Check out the
Standard GGUF Sister Repository for uncalibrated and full 8-bit precision options.
🎯 Matrix-Weighted Calibration (Imatrix)
An Importance Matrix (imatrix) calculation tracks activations across network layers using a calibration sequence, then weights the quantization process to preserve the parameters that matter most for output quality — improving fidelity at low bit depths.
➡️ Calibration dataset: Bartowski's calibration_datav5.txt.
[!NOTE]
IQ4_NL is included because the matrix enables a non-linear 4-bit format that outperforms standard linear 4-bit quantization.Q8_0 is absent because 8-bit quantization already introduces near-zero degradation, making calibration unnecessary — see the standard sister repository for that variant.
ℹ️ Model Profile & Core Features
LFM2.5-VL-450M is the smaller vision-language model in Liquid AI's LFM2.5-VL family, pairing a compact LFM2 hybrid language backbone with a SigLIP2 NaFlex vision encoder for lightweight image understanding on constrained hardware. It targets edge and mobile deployment scenarios where the larger 1.6B variant would be impractical, while retaining single- and multi-image support and a 32,768-token context window.
The
heretic suffix denotes post-processing via the
Heretic v1.3.0 method performed by
coder3101, which suppresses refusal behavior while preserving the model's vision-language capabilities.
📋 Technical Specifications
| Property | Value |
|---|
| Base Architecture | LFM2 hybrid (gated conv + GQA) + SigLIP2 NaFlex vision encoder |
| Developed by | Liquid AI |
| Total Parameters | 450M (LM + vision encoder) |
| Vision Encoder | SigLIP2 NaFlex, shape-optimized |
| Primary Use | Lightweight image understanding, edge/mobile deployment |
| Context Window | 32,768 tokens |
| Vision Projector | Integrated (see repository files below) |
| Languages | English, Arabic, Chinese, French, German, Japanese, Korean, Spanish, Portuguese |
| Abliteration Tool | Heretic v1.3.0 |
| Prompt Format | ChatML |
🛠️ Heretic Overrides (ARA)
| Property | Value |
|---|
| direction_index | 10.86 |
| attn.o_proj.max_weight | 1.01 |
| attn.o_proj.max_weight_position | 9.18 |
| attn.o_proj.min_weight | 0.37 |
| attn.o_proj.min_weight_distance | 5.15 |
| mlp.down_proj.max_weight | 1.18 |
| mlp.down_proj.max_weight_position | 12.27 |
| mlp.down_proj.min_weight | 0.40 |
| mlp.down_proj.min_weight_distance | 4.19 |
📊 Refusal Bypass Metrics
[!NOTE]
The metrics below are self-reported by the original model author (
coder3101) and have not been independently reproduced.
🧮 Numerical & Tensor Formats
| Property | Value |
|---|
| Text Tensor Types | IQ4_NL, Q4_K_M, Q5_K_M (all with imatrix calibration) |
| Importance Matrix | Bartowski's calibration_datav5.txt |
| Vision Tensors | Q8_0, BF16 |
📦 Available Model Files
Main model weights
| Filename | Quantization | llama.cpp Build | Size | Download |
|---|
LFM2.5-VL-450M-heretic-IQ4_NL-imatrix.gguf | IQ4_NL | b9860 | 209 MB | 📥 Download |
LFM2.5-VL-450M-heretic-Q4_K_M-imatrix.gguf | Q4_K_M | b9860 | 219 MB | 📥 Download |
LFM2.5-VL-450M-heretic-Q5_K_M-imatrix.gguf | Q5_K_M | b9860 | 248 MB | 📥 Download |
mmproj — vision projector files
| Filename | Quantization | Size | Download |
|---|
mmproj-LFM2.5-VL-450M-heretic-Q8_0.gguf | Q8_0 | 98 MB | 📥 Download |
mmproj-LFM2.5-VL-450M-heretic-BF16.gguf | BF16 | 181 MB | 📥 Download |
🎛️ Component Pairing Guide
Download exactly one main weights file:
IQ4_NL: Non-linear 4-bit format, best choice for constrained memory when imatrix calibration is present.Q4_K_M: Balanced 4-bit format suitable for most everyday use.Q5_K_M: Higher-fidelity mid-range format recommended as a general default.
mmproj files (optional): multimodal vision projectors. Pass one via the --mmproj flag in llama.cpp to enable image input.
BF16 (Recommended): Highest possible image processing accuracy. While older projectors were small, modern vision towers can hover around 1GB. If you are tight on VRAM, it is completely viable to run this on system RAM (CPU) with a minimal performance penalty, saving your precious GPU space for the main model layers.Q8_0: Cuts the projector file size and memory footprint in half (~500MB for larger 1GB files). Use this if you prefer to keep the vision tower hosted entirely on your GPU but need to claw back some VRAM to avoid Out-Of-Memory (OOM) crashes.
⚡ Deployment & Execution Commands
[!IMPORTANT]
The vision projector (--mmproj) must be supplied at runtime whenever image inputs are used. Omitting it disables multimodal capability entirely.
[!NOTE]
LiquidAI recommends the following sampling configuration for best results:
- Text:
temperature=0.1, min_p=0.15, repetition_penalty=1.05.
- Vision:
min_image_tokens=32, max_image_tokens=256, do_image_splitting=True.
[!TIP]
Swap the -m filename below for either quantized file depending on your size/quality trade-off preference.
llama.cpp CLI (with image)
1./llama-cli \
2 -m LFM2.5-VL-450M-heretic-IQ4_NL-imatrix.gguf \
3 --mmproj mmproj-LFM2.5-VL-450M-heretic-Q8_0.gguf \
4 -c 8192 \
5 -ngl 99 \
6 --image "path/to/image.jpg" \
7 -p "<|im_start|>user\nDescribe what you see in this image.<|im_end|>\n<|im_start|>assistant\n"
OpenAI-Compatible API Server
1./llama-server \
2 --host 0.0.0.0 \
3 --port 8080 \
4 -m LFM2.5-VL-450M-heretic-IQ4_NL-imatrix.gguf \
5 --mmproj mmproj-LFM2.5-VL-450M-heretic-Q8_0.gguf \
6 -c 16384 \
7 -ngl 99 \
8 --flash-attn
💬 Chat Templates & Prompt Design (ChatML)
1<|im_start|>system
2You are a helpful multimodal assistant.<|im_end|>
3<|im_start|>user
4Your question or image payload here.<|im_end|>
5<|im_start|>assistant
⚠️ Safety & Operational Notes
- This model is abliterated and will generate content that standard aligned models refuse. Use responsibly and in compliance with applicable laws.
- Fits comfortably on a single GPU with at least 8 GB VRAM at quantized precision.
- Context window is limited to 32,768 tokens — shorter than the text-only LFM2.5 models.
- Vision tensors are kept at BF16 or Q8_0 depending on the mmproj variant chosen, to preserve visual feature quality.
- Imatrix calibration improves perplexity recovery compared to non-imatrix quantization, particularly on low-frequency tokens.
- IQ4_NL produces a smaller file than Q4_K_M and tends to run faster on CPU and ARM devices; imatrix calibration narrows the quality gap between the two formats considerably.
