Up to date
232M parameters, images and text in, text out, 1K tokens of context.
Verified 2026-10-25 Rank #509 in LLMs · score 76.2 method v1 Sources 3 Downloads 30d 176.7k · #262 of 619 tracked
Changed this week: 2 new quantized builds on the Hub (9 in all).2026-10-24
Picked by rule from 5 tracked pieces: the lab's own post, the most shared, a field report, the latest.
Microsoft released Florence 2, a model capable of image captioning, segmentation, and OCR that runs on CPU and GPU.
The 1 pick above, plus these 4. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.
Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.
No API provider serves it, so its downloads are people running it themselves: 9 quantized builds and 100 Spaces.
Top Spaces running it: gokaygokay/Florence-2 ♥ 848 · SkalskiP/better-florence-2 ♥ 198 · SkalskiP/florence-2-video ♥ 72 · gokaygokay/Florence-2-SD3-Captioner ♥ 39
Weights only, an estimate: leave room for the KV cache. Method
About 0.1 GB at INT4, 0.3 GB at FP8, 0.6 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
Weights, every dated article, and repo commits on one rail.
Teal is the weights (2024-06-15), amber triangles are 5 dated articles (each links to it), grey ticks are repo commits.
52 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.
Everything else, folded: the full tables, every source.
| Benchmark | Value | Setting / notes | Reported by | Date |
|---|---|---|---|---|
| author-reported COCO Caption Karpathy test CIDEr | 140 | Florence-2-base-ft | model author | 2026-10-07 |
| author-reported NoCaps val CIDEr | 116.7 | Florence-2-base-ft | model author | 2026-10-07 |
| author-reported TextCaps val CIDEr | 143.9 | Florence-2-base-ft | model author | 2026-10-07 |
| author-reported TextVQA · Understand images and video | 63.6 percent | Florence-2-base-ft | model author | 2026-10-07 |
| author-reported VQAv2 · Understand images and video | 79.7 percent | Florence-2-base-ft | model author | 2026-10-07 |
| author-reported VizWiz VQA test-dev Acc | 63.6 percent | Florence-2-base-ft | model author | 2026-10-07 |
This fine-tuned model uses a sequence-to-sequence architecture to perform vision tasks like captioning and object detection.[1]
Generated 2026-10-07 15:13 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.
Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.
Listed in LLMs
| Pipeline tag | image-text-to-text |
|---|---|
| Library | transformers |
| Architectures | Florence2ForConditionalGeneration |
| Model type | florence2 |
| Dtype | float16 |
| Params by dtype | F16: 231,567,705 |
| Size band | S |
| MoE | no |
| Context length | 1,024 (config.json max_position_embeddings) |
| Papers | arXiv:2311.06242 |
| License | MIT (mit) |
| Hub snapshot | 2026-10-25 |
vLLM: not in registry · Florence2ForConditionalGeneration · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: not in registry · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed
vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.
Weights-only memory for Florence-2-base-ft: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 0.6 GB · Edge (<4 GB) |
|---|---|
| FP8 (1 byte/param) | 0.3 GB · Edge (<4 GB) |
| INT4 (0.5 bytes/param) | 0.1 GB · Edge (<4 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
finetune: 23quantized: 9adapter: 19merge: 1
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| aipib/Florence-2-VQAJP2 | finetune | — | 271M | S | mit | aipib | 1 | 429.5k |
| onnx-community/Florence-2-base-ft | quantized | — | — | — | mit | onnx-community | 44 | 408.7k |
| HuggingFaceM4/Florence-2-DocVQA | finetune | — | 823M | S | — | HuggingFaceM4 | 65 | 50.7k |
| Vimax97/Florence-2-base-gpt4_captioner_v1 | finetune | — | 271M | S | mit | Vimax97 | 4 | 5.8k |
| amayuelas/plot-visualization-florence-2-lora-32 | adapter | — | — | — | mit | amayuelas | 1 | 4.6k |
| nazounoryuu/florence_base__mixed__page__line_od | finetune | — | 271M | S | mit | nazounoryuu | 0 | 2.8k |
| prithivMLmods/Florence-2-VLM-Doc-VQA | finetune | — | 271M | S | — | prithivMLmods | 5 | 2.4k |
| remyxai/SpaceFlorence-2 | finetune | — | 271M | S | mit | remyxai | 2 | 2.2k |
| byh711/FLODA-deepfake | adapter | — | — | — | apache-2.0 | byh711 | 0 | 1.7k |
| devxyasir/florence-finetuned-license-plate-detection | finetune | — | 271M | S | mit | devxyasir | 1 | 1.6k |
| happyme531/Florence-2-base-ft-ONNX-RKNN2 | quantized | — | — | — | agpl-3.0 | happyme531 | 10 | 521 |
| adamchanadam/Test_Florence-2-FT-DocVQA | finetune | — | 271M | S | mit | adamchanadam | 0 | 1k |
| Proteus-Computer-Use/omniparser-finetuned | finetune | — | 232M | S | mit | Proteus-Computer-Use | 4 | 624 |
| aipib/Florence-2-OCRJP-remake4 | finetune | — | 271M | S | mit | aipib | 0 | 708 |
| remyxai/PoseFlorence-2 | finetune | — | 271M | S | — | remyxai | 3 | 506 |
| nirusanan/Florence-2_FT_Lung-Cancer-detection | finetune | — | 271M | S | — | nirusanan | 1 | 487 |
| aipib/Florence-2-FT-JP-OCR2 | finetune | — | 271M | S | mit | aipib | 0 | 429 |
| nazounoryuu/florence_base__mixed__line_bbox__ocr | finetune | — | 271M | S | mit | nazounoryuu | 0 | 419 |
| nirusanan/Florence-2_FT_VQA | finetune | — | 271M | S | — | nirusanan | 0 | 360 |
| Heliosoph/florence-2-base-ft-quantized-onnx | quantized | -quant | — | — | mit | Heliosoph | 0 | 324 |
| byh711/Florence-TableDetection | finetune | — | 823M | S | mit | byh711 | 0 | 219 |
| topguy/Florence-2-base-ft-compat | finetune | — | 232M | S | apache-2.0 | topguy | 0 | 164 |
| byh711/Florence2-Problem-detection | finetune | — | 271M | S | mit | byh711 | 0 | 158 |
| aghoraguru/florence2_lora_vqa | adapter | — | — | — | mit | aghoraguru | 0 | 155 |
| byh711/Florence2-Table-detection | finetune | — | 271M | S | mit | byh711 | 0 | 127 |
| Alasil/Classification_NU | adapter | — | 232M | S | — | Alasil | 0 | 120 |
| evrcc7881/utility-pole-florenceLoRA | adapter | — | — | — | — | evrcc7881 | 0 | 107 |
| minhvn4/florence2-wave-ui-lora | adapter | — | 271M | S | mit | minhvn4 | 0 | 84 |
| chungvodim/florence-2-vehicle-damage | adapter | — | — | — | — | chungvodim | 0 | 81 |
| keyi10/florence2_finetuning_example | adapter | — | — | — | — | keyi10 | 0 | 73 |
| controlllmai/Florence-2-base-ft | quantized | — | — | — | mit | controlllmai | 0 | 71 |
| arvisioncode/florence_50_5e6_2k | adapter | — | — | — | — | arvisioncode | 0 | 70 |
| Heliosoph/florence-2-base-ft-fp16-onnx | quantized | -fp16 | — | — | mit | Heliosoph | 0 | 66 |
| textagent/Florence-2-base-ft | quantized | — | — | — | mit | textagent | 0 | 53 |
| aniketmaurya/receipt-model-2025 | finetune | — | — | — | mit | aniketmaurya | 1 | 0 |
| kukiui/electricFlorence | adapter | — | 232M | S | mit | kukiui | 0 | 48 |
| Khabner/florence-base-lora-v22 | adapter | — | — | — | mit | Khabner | 0 | 48 |
| vlad-grig/Florence-2-base-ft | quantized | — | — | — | mit | vlad-grig | 0 | 46 |
| sageofai/Florence2-MEDVQA-GI-2026-Task1 | finetune | — | 232M | S | cc-by-nc-4.0 | sageofai | 0 | 41 |
| Khabner/florence-base-lora-v1 | adapter | — | — | — | mit | Khabner | 0 | 39 |
| Date | Commit | SHA |
|---|---|---|
| 2025-08-04 | Fix weight init, deprecation warnings, safetensors (#29) | f6c1a25 |
| 2024-07-20 | update_model_init_fp16 (#20) | 9803f52 |
| 2024-07-19 | update_ocr_postprocess (#19) | f2be9b1 |
| 2024-07-01 | Update pipeline tag | ace966b |
| 2024-06-30 | Fix modeling code (typos/bugs) (#6) | 56dfc91 |
| 2024-06-26 | Fix incorrect bos_token, eos_token, and pad_token ids in config.json (#7) | 9cef882 |
| 2024-06-18 | Update README.md | e7a5acc |
| 2024-06-18 | update_readme_example (#3) | e4e3ff8 |
| 2024-06-18 | add_grounding_example (#2) | 5117f5a |
| 2024-06-15 | init_model (#1) | 494b983 |
| 2024-06-15 | initial commit | 17b32c0 |
Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.
arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).
Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:41 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:46 UTC · Summary: 2026-10-07 15:13 UTC · Last verified 2026-10-25
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.
