Technical report: Read paper · PDF · BibTeX · All models
Collection: LFM vision-language
Usage: Step-by-step Windows, Linux and macOS instructions.
LFM3-VL 2B Heretic — BF16
LFM3-VL-2B Heretic checkpoint in Hugging Face Safetensors format, with a language model, vision encoder, projector, tokenizer, processor configuration, and custom model code.
| File | Size (decimal GB) |
|---|---|
model.safetensors |
3.836 |
Format and compatibility
The configuration declares Lfm3VlForConditionalGeneration, BF16 weights, and Transformers 5.21.0. Use a compatible Transformers release and the model’s image processor and chat template.
Load
from transformers import AutoModelForImageTextToText, AutoProcessor
repo = "Lucasgitguy-co/lfm3-vl-2b-heretic"
processor = AutoProcessor.from_pretrained(repo, trust_remote_code=True)
model = AutoModelForImageTextToText.from_pretrained(
repo, torch_dtype="auto", trust_remote_code=True
)
The custom-code loading option executes repository Python code. The example initializes the model; image preprocessing and generation should follow the upstream LFM3-VL guide.
Related formats
For upstream model information and license terms, see LiquidAI/LFM3-VL-2B.
Evaluation and provenance
This card describes the repository files and configuration. Detailed training data, training procedure, and benchmark results are not documented in the inspected artifacts. Loading examples are guidance, not a claim of measured model quality.
License and attribution
The model weights, derived quantizations, and upstream model implementation remain subject to the LFM Open License v1.0 in LICENSE. Its commercial-use limitation applies to derivatives; this release is not an Apache-2.0 model. The publisher’s original documentation additions are available under Apache 2.0, without changing the terms of upstream material.