lfm2_vl
Datasets
All datasets matching “lfm2_vl”AI-Bicycle-LFM2-VL-450M
🧠 Multistage Japanese Dialogue & VQA Dataset
本データセットは、複数の公開データセットを統合・日本語化し、
自然な日本語対話および視覚応答（Vision Reaction）データを生成したものです。
出典を明示することで、商用利用が可能です。
ライセンスを尊重したうえで翻訳・加工を行い、3つのステージで構成されています。
注意: Vision Reaction とは、画像に対して自然な応答をすることです。
例えば、海の画像を入力したとき、一般的なVQAタスクでは「青い空と青い海が広がっています。....」のような説明をします。
Vision Reactionでは、「お、きれいな海だな！」というように、自然なリアクションをします。
📘 データ構成
Stage 1：en_multiturn.jsonl
元データ：allenai/soda
ライセンス：Creative Commons Attribution 4.0 International (CC BY 4.0)… See the full description on the dataset page: https://huggingface.co/datasets/HayatoHongo/AI-Bicycle-LFM2-VL-450M.lfm2vl-lab-results
lfm2vl-lab — raw evaluation results
Every number behind lfm2vl-lab, a
component-wise quantization and edge-inference study of
LiquidAI/LFM2.5-VL-3B on Apple Silicon
(M4 Pro, 24 GB).
Published separately from the code so the numbers stay citable and diffable on their own.
Weights in that study are ephemeral by design — converted, evaluated, recorded, deleted.
These JSON files are what survives.
Files
file
what it holds
coverage_audit_gs64.json
per-module… See the full description on the dataset page: https://huggingface.co/datasets/shubhambaid/lfm2vl-lab-results.lfm2-vl-extract-smoke-test
Structured extraction with LFM2.5-VL-1.6B-Extract
JSON fields extracted from images in davanstrien/ufo-ColPali
using LiquidAI/LFM2.5-VL-1.6B-Extract.
Source: davanstrien/ufo-ColPali (split train)
Model: LiquidAI/LFM2.5-VL-1.6B-Extract
Fields: title, date, summary
Output column: extraction (JSON string per row)
Valid JSON: 5/5
Date: 2026-06-05 15:54 UTC
Generated with the uv-scripts/ocr lfm2-vl-extract.py script.
lfm2-vl-extract-smoke2
Structured extraction with LFM2.5-VL-1.6B-Extract
JSON fields extracted from images in davanstrien/ufo-ColPali
using LiquidAI/LFM2.5-VL-1.6B-Extract.
Source: davanstrien/ufo-ColPali (split train)
Model: LiquidAI/LFM2.5-VL-1.6B-Extract
Fields: document_type, date, summary
Output column: extraction (JSON string per row)
Valid JSON: 2/2
Date: 2026-06-05 08:54 UTC
Generated with the uv-scripts/ocr lfm2-vl-extract.py script.
lfm2-vl-extract-smoke
Structured extraction with LFM2.5-VL-1.6B-Extract
JSON fields extracted from images in davanstrien/ufo-ColPali
using LiquidAI/LFM2.5-VL-1.6B-Extract.
Source: davanstrien/ufo-ColPali (split train)
Model: LiquidAI/LFM2.5-VL-1.6B-Extract
Fields: document_type, date, summary
Output column: extraction (JSON string per row)
Valid JSON: 2/2
Date: 2026-06-05 08:44 UTC
Generated with the uv-scripts/ocr lfm2-vl-extract.py script.
lfm2-vl-extract-cord-test
Structured extraction with LFM2.5-VL-1.6B-Extract
JSON fields extracted from images in naver-clova-ix/cord-v2
using LiquidAI/LFM2.5-VL-1.6B-Extract.
Source: naver-clova-ix/cord-v2 (split train)
Model: LiquidAI/LFM2.5-VL-1.6B-Extract
Fields: store_name, date, total
Output column: extraction (JSON string per row)
Valid JSON: 5/5
Date: 2026-06-05 15:59 UTC
Generated with the uv-scripts/ocr lfm2-vl-extract.py script.