inria-soda/carte-benchmark
Updated
•
90
•
3
SODA
non-profit
AI & ML interests
Computational and statistical research in data science and machine learning, basic and applied, to harness large databases on health and society.
Recent Activity
View all activity
inria-soda's activity
junkim930
updated
a
dataset
about 24 hours ago
Post
3496
supercharge your LLM apps with smolagents 🔥
however cool your LLM is, without being agentic it can only go so far
enter smolagents: a new agent library by Hugging Face to make the LLM write code, do analysis and automate boring stuff!
Here's our blog for you to get started https://huggingface.co/blog/smolagents
however cool your LLM is, without being agentic it can only go so far
enter smolagents: a new agent library by Hugging Face to make the LLM write code, do analysis and automate boring stuff!
Here's our blog for you to get started https://huggingface.co/blog/smolagents
Post
4095
QwQ can see 🔥
Qwen team released QvQ, a large vision LM with reasoning 😱
it outperforms proprietary VLMs on several benchmarks, comes with open weights and a demo!
Check them out ⬇️
Demo Qwen/QVQ-72B-preview
Model Qwen/QVQ-72B-Preview
Read more https://qwenlm.github.io/blog/qvq-72b-preview/
Congratulations @JustinLin610 and team!
Qwen team released QvQ, a large vision LM with reasoning 😱
it outperforms proprietary VLMs on several benchmarks, comes with open weights and a demo!
Check them out ⬇️
Demo Qwen/QVQ-72B-preview
Model Qwen/QVQ-72B-Preview
Read more https://qwenlm.github.io/blog/qvq-72b-preview/
Congratulations @JustinLin610 and team!
- 2 replies
Post
2723
Aya by Cohere For AI can now see! 👀
C4AI community has built Maya 8B, a new open-source multilingual VLM built on SigLIP and Aya 8B 🌱 works on 8 languages! 🗣️
The authors extend Llava dataset using Aya's translation capabilities with 558k examples!
ry it here kkr5155/maya_demo
Dataset maya-multimodal/pretrain
Model maya-multimodal/maya 👏
kudos @nahidalam and team
C4AI community has built Maya 8B, a new open-source multilingual VLM built on SigLIP and Aya 8B 🌱 works on 8 languages! 🗣️
The authors extend Llava dataset using Aya's translation capabilities with 558k examples!
ry it here kkr5155/maya_demo
Dataset maya-multimodal/pretrain
Model maya-multimodal/maya 👏
kudos @nahidalam and team
- 1 reply
Post
3153
Apollo is a new family of open-source video language models by Meta, where 3B model outperforms most 7B models and 7B outperforms most 30B models 🧶
✨ the models come in 1.5B https://huggingface.co/Apollo-LMMs/Apollo-1_5B-t32, 3B https://huggingface.co/Apollo-LMMs/Apollo-3B-t32 and 7B https://huggingface.co/Apollo-LMMs/Apollo-7B-t32 with A2.0 license, based on Qwen1.5 & Qwen2
✨ the authors also release a benchmark dataset https://huggingface.co/spaces/Apollo-LMMs/ApolloBench
The paper has a lot of experiments (they trained 84 models!) about what makes the video LMs work ⏯️
Try the demo for best setup here https://huggingface.co/spaces/Apollo-LMMs/Apollo-3B
they evaluate sampling strategies, scaling laws for models and datasets, video representation and more!
> The authors find out that whatever design decision was applied to small models also scale properly when the model and dataset are scaled 📈 scaling dataset has diminishing returns for smaller models
> They evaluate frame sampling strategies, and find that FPS sampling is better than uniform sampling, and they find 8-32 tokens per frame optimal
> They also compare image encoders, they try a variation of models from shape optimized SigLIP to DINOv2
they find google/siglip-so400m-patch14-384 to be most powerful 🔥
> they also compare freezing different parts of models, training all stages with some frozen parts give the best yield
They eventually release three models, where Apollo-3B outperforms most 7B models and Apollo 7B outperforms 30B models 🔥
✨ the models come in 1.5B https://huggingface.co/Apollo-LMMs/Apollo-1_5B-t32, 3B https://huggingface.co/Apollo-LMMs/Apollo-3B-t32 and 7B https://huggingface.co/Apollo-LMMs/Apollo-7B-t32 with A2.0 license, based on Qwen1.5 & Qwen2
✨ the authors also release a benchmark dataset https://huggingface.co/spaces/Apollo-LMMs/ApolloBench
The paper has a lot of experiments (they trained 84 models!) about what makes the video LMs work ⏯️
Try the demo for best setup here https://huggingface.co/spaces/Apollo-LMMs/Apollo-3B
they evaluate sampling strategies, scaling laws for models and datasets, video representation and more!
> The authors find out that whatever design decision was applied to small models also scale properly when the model and dataset are scaled 📈 scaling dataset has diminishing returns for smaller models
> They evaluate frame sampling strategies, and find that FPS sampling is better than uniform sampling, and they find 8-32 tokens per frame optimal
> They also compare image encoders, they try a variation of models from shape optimized SigLIP to DINOv2
they find google/siglip-so400m-patch14-384 to be most powerful 🔥
> they also compare freezing different parts of models, training all stages with some frozen parts give the best yield
They eventually release three models, where Apollo-3B outperforms most 7B models and Apollo 7B outperforms 30B models 🔥
- 6 replies
Post
1735
A complete RAG pipeline includes a reranker, which ranks the documents to find the best document 📓
Same goes for multimodal RAG, multimodal rerankers which we can integrate to multimodal RAG pipelines!
Learn how to build a complete multimodal RAG pipeline with vidore/colqwen2-v1.0 as retriever, lightonai/MonoQwen2-VL-v0.1 as reranker, Qwen/Qwen2-VL-7B-Instruct as VLM in this notebook that runs on a GPU as small as L4 🔥 https://huggingface.co/learn/cookbook/multimodal_rag_using_document_retrieval_and_reranker_and_vlms
Same goes for multimodal RAG, multimodal rerankers which we can integrate to multimodal RAG pipelines!
Learn how to build a complete multimodal RAG pipeline with vidore/colqwen2-v1.0 as retriever, lightonai/MonoQwen2-VL-v0.1 as reranker, Qwen/Qwen2-VL-7B-Instruct as VLM in this notebook that runs on a GPU as small as L4 🔥 https://huggingface.co/learn/cookbook/multimodal_rag_using_document_retrieval_and_reranker_and_vlms
Post
5557
This week in open-source AI was insane 🤠 A small recap🕺🏻
merve/dec-6-releases-67545caebe9fc4776faac0a3
Multimodal 🖼️
> Google shipped a PaliGemma 2, new iteration of PaliGemma with more sizes: 3B, 10B and 28B, with pre-trained and captioning variants 👏
> OpenGVLab released InternVL2, seven new vision LMs in different sizes, with sota checkpoint with MIT license ✨
> Qwen team at Alibaba released the base models of Qwen2VL models with 2B, 7B and 72B ckpts
LLMs 💬
> Meta released a new iteration of Llama 70B, Llama3.2-70B trained further
> EuroLLM-9B-Instruct is a new multilingual LLM for European languages with Apache 2.0 license 🔥
> Dataset: CohereForAI released GlobalMMLU, multilingual version of MMLU with 42 languages with Apache 2.0 license
> Dataset: QwQ-LongCoT-130K is a new dataset to train reasoning models
> Dataset: FineWeb2 just landed with multilinguality update! 🔥 nearly 8TB pretraining data in many languages!
Image/Video Generation 🖼️
> Tencent released HunyuanVideo, a new photorealistic video generation model
> OminiControl is a new editing/control framework for image generation models like Flux
Audio 🔊
> Indic-Parler-TTS is a new text2speech model made by community
Multimodal 🖼️
> Google shipped a PaliGemma 2, new iteration of PaliGemma with more sizes: 3B, 10B and 28B, with pre-trained and captioning variants 👏
> OpenGVLab released InternVL2, seven new vision LMs in different sizes, with sota checkpoint with MIT license ✨
> Qwen team at Alibaba released the base models of Qwen2VL models with 2B, 7B and 72B ckpts
LLMs 💬
> Meta released a new iteration of Llama 70B, Llama3.2-70B trained further
> EuroLLM-9B-Instruct is a new multilingual LLM for European languages with Apache 2.0 license 🔥
> Dataset: CohereForAI released GlobalMMLU, multilingual version of MMLU with 42 languages with Apache 2.0 license
> Dataset: QwQ-LongCoT-130K is a new dataset to train reasoning models
> Dataset: FineWeb2 just landed with multilinguality update! 🔥 nearly 8TB pretraining data in many languages!
Image/Video Generation 🖼️
> Tencent released HunyuanVideo, a new photorealistic video generation model
> OminiControl is a new editing/control framework for image generation models like Flux
Audio 🔊
> Indic-Parler-TTS is a new text2speech model made by community
Post
1532
New InternVL drop with a state-of-the-art 78B vision language model with MIT license 🔥 https://huggingface.co/collections/OpenGVLab/internvl-25-673e1019b66e2218f68d7c1c
The release comes with seven new vision LMs based on InternViT 300M/6B and Qwen2.5 (0.5B, 3B, 32B, 72B) and InternLM2 (8B, 7B, 20B) in different sizes
78B model is of InternViT 6B and Qwen2.5-72B Instruct, can accomplish variety of tasks 👏 Try here OpenGVLab/InternVL
The release comes with seven new vision LMs based on InternViT 300M/6B and Qwen2.5 (0.5B, 3B, 32B, 72B) and InternLM2 (8B, 7B, 20B) in different sizes
78B model is of InternViT 6B and Qwen2.5-72B Instruct, can accomplish variety of tasks 👏 Try here OpenGVLab/InternVL
Post
2647
small but mighty 🔥
you can fine-tune SmolVLM on an L4 with batch size of 4 and it will only take 16.4 GB VRAM 🫰🏻 also with gradient accumulation simulated batch size is 16 ✨
I made a notebook that includes all the goodies: QLoRA, gradient accumulation, gradient checkpointing with explanations on how they work 💝 https://github.com/huggingface/smollm/blob/main/finetuning/Smol_VLM_FT.ipynb
you can fine-tune SmolVLM on an L4 with batch size of 4 and it will only take 16.4 GB VRAM 🫰🏻 also with gradient accumulation simulated batch size is 16 ✨
I made a notebook that includes all the goodies: QLoRA, gradient accumulation, gradient checkpointing with explanations on how they work 💝 https://github.com/huggingface/smollm/blob/main/finetuning/Smol_VLM_FT.ipynb
Post
2889
Last week we were blessed with open-source models! A recap 💝
merve/nov-29-releases-674ccc255a57baf97b1e2d31
🖼️ Multimodal
> At Hugging Face we released SmolVLM, a performant and efficient smol vision language model 💗
> Show Lab released ShowUI-2B: new vision-language-action model to build GUI/web automation agents 🤖
> Rhymes AI has released the base model of Aria: Aria-Base-64K and Aria-Base-8K with their respective context length
> ViDoRe team released ColSmolVLM: A new ColPali-like retrieval model based on SmolVLM
> Dataset: Llava-CoT-o1-Instruct: new dataset labelled using Llava-CoT multimodal reasoning model📖
> Dataset: LLaVA-CoT-100k dataset used to train Llava-CoT released by creators of Llava-CoT 📕
💬 LLMs
> Qwen team released QwQ-32B-Preview, state-of-the-art open-source reasoning model, broke the internet 🔥
> AliBaba has released Marco-o1, a new open-source reasoning model 💥
> NVIDIA released Hymba 1.5B Base and Instruct, the new state-of-the-art SLMs with hybrid architecture (Mamba + transformer)
⏯️ Image/Video Generation
> Qwen2VL-Flux: new image generation model based on Qwen2VL image encoder, T5 and Flux for generation
> Lightricks released LTX-Video, a new DiT-based video generation model that can generate 24 FPS videos at 768x512 res ⏯️
> Dataset: Image Preferences is a new image generation preference dataset made with DIBT community effort of Argilla 🏷️
Audio
> OuteAI released OuteTTS-0.2-500M new multilingual text-to-speech model based on Qwen-2.5-0.5B trained on 5B audio prompt tokens
merve/nov-29-releases-674ccc255a57baf97b1e2d31
🖼️ Multimodal
> At Hugging Face we released SmolVLM, a performant and efficient smol vision language model 💗
> Show Lab released ShowUI-2B: new vision-language-action model to build GUI/web automation agents 🤖
> Rhymes AI has released the base model of Aria: Aria-Base-64K and Aria-Base-8K with their respective context length
> ViDoRe team released ColSmolVLM: A new ColPali-like retrieval model based on SmolVLM
> Dataset: Llava-CoT-o1-Instruct: new dataset labelled using Llava-CoT multimodal reasoning model📖
> Dataset: LLaVA-CoT-100k dataset used to train Llava-CoT released by creators of Llava-CoT 📕
💬 LLMs
> Qwen team released QwQ-32B-Preview, state-of-the-art open-source reasoning model, broke the internet 🔥
> AliBaba has released Marco-o1, a new open-source reasoning model 💥
> NVIDIA released Hymba 1.5B Base and Instruct, the new state-of-the-art SLMs with hybrid architecture (Mamba + transformer)
⏯️ Image/Video Generation
> Qwen2VL-Flux: new image generation model based on Qwen2VL image encoder, T5 and Flux for generation
> Lightricks released LTX-Video, a new DiT-based video generation model that can generate 24 FPS videos at 768x512 res ⏯️
> Dataset: Image Preferences is a new image generation preference dataset made with DIBT community effort of Argilla 🏷️
Audio
> OuteAI released OuteTTS-0.2-500M new multilingual text-to-speech model based on Qwen-2.5-0.5B trained on 5B audio prompt tokens
Post
2178
The authors of ColPali trained a retrieval model based on SmolVLM 🤠
vidore/colsmolvlm-alpha
TLDR;
- ColSmolVLM performs better than ColPali and DSE-Qwen2 on all English tasks
- ColSmolVLM is more memory efficient than ColQwen2 💗
TLDR;
- ColSmolVLM performs better than ColPali and DSE-Qwen2 on all English tasks
- ColSmolVLM is more memory efficient than ColQwen2 💗
Post
3897
Small yet mighty! 💫
We are releasing SmolVLM: a new 2B small vision language made for on-device use, fine-tunable on consumer GPU, immensely memory efficient 🤠
We release three checkpoints under Apache 2.0: SmolVLM-Instruct, SmolVLM-Synthetic and SmolVLM-Base HuggingFaceTB/smolvlm-6740bd584b2dcbf51ecb1f39
Learn more from our blog here: huggingface.co/blog/smolvlm
This release comes with a demo, fine-tuning code, MLX integration and TRL integration for DPO 💝
Try the demo: HuggingFaceTB/SmolVLM
Fine-tuning Recipe: https://github.com/huggingface/smollm/blob/main/finetuning/Smol_VLM_FT.ipynb
Also TRL integration for DPO 💗
We are releasing SmolVLM: a new 2B small vision language made for on-device use, fine-tunable on consumer GPU, immensely memory efficient 🤠
We release three checkpoints under Apache 2.0: SmolVLM-Instruct, SmolVLM-Synthetic and SmolVLM-Base HuggingFaceTB/smolvlm-6740bd584b2dcbf51ecb1f39
Learn more from our blog here: huggingface.co/blog/smolvlm
This release comes with a demo, fine-tuning code, MLX integration and TRL integration for DPO 💝
Try the demo: HuggingFaceTB/SmolVLM
Fine-tuning Recipe: https://github.com/huggingface/smollm/blob/main/finetuning/Smol_VLM_FT.ipynb
Also TRL integration for DPO 💗
Post
2593
What a week! A recap for everything you missed ❄️
merve/nov-22-releases-673fbbcfc1c97c4f411def07
Multimodal ✨
> Mistral AI
released Pixtral 124B, a gigantic open vision language model
> Llava-CoT (formerly known as Llava-o1) was released, a multimodal reproduction of o1 model by PKU
> OpenGVLab released MMPR: a new multimodal reasoning dataset
> Jina has released Jina-CLIP-v2 0.98B multilingual multimodal embeddings
> Apple released new SotA vision encoders AIMv2
LLMs 🦙
> AllenAI dropped a huge release of models, datasets and scripts for Tülu, a family of models based on Llama 3.1 aligned with SFT, DPO and a new technique they have developed called RLVR
> Jina has released embeddings-v3: new multilingual embeddings with longer context
> Hugging Face released SmolTalk: synthetic dataset used to align SmolLM2 using supervised fine-tuning
> Microsoft released orca-agentinstruct-1M-v1: a gigantic instruction dataset of 1M synthetic instruction pairs
Image Generation 🖼️
> Black Forest Labs released Flux 1. tools: four new models for different image modifications and two LoRAs to do image conditioning and better steer generations
Lastly Hugging Face released a new library Observers: a lightweight SDK for monitoring interactions with AI APIs and easily store and browse them 📚
$ pip install observers
merve/nov-22-releases-673fbbcfc1c97c4f411def07
Multimodal ✨
> Mistral AI
released Pixtral 124B, a gigantic open vision language model
> Llava-CoT (formerly known as Llava-o1) was released, a multimodal reproduction of o1 model by PKU
> OpenGVLab released MMPR: a new multimodal reasoning dataset
> Jina has released Jina-CLIP-v2 0.98B multilingual multimodal embeddings
> Apple released new SotA vision encoders AIMv2
LLMs 🦙
> AllenAI dropped a huge release of models, datasets and scripts for Tülu, a family of models based on Llama 3.1 aligned with SFT, DPO and a new technique they have developed called RLVR
> Jina has released embeddings-v3: new multilingual embeddings with longer context
> Hugging Face released SmolTalk: synthetic dataset used to align SmolLM2 using supervised fine-tuning
> Microsoft released orca-agentinstruct-1M-v1: a gigantic instruction dataset of 1M synthetic instruction pairs
Image Generation 🖼️
> Black Forest Labs released Flux 1. tools: four new models for different image modifications and two LoRAs to do image conditioning and better steer generations
Lastly Hugging Face released a new library Observers: a lightweight SDK for monitoring interactions with AI APIs and easily store and browse them 📚
$ pip install observers
- 3 replies
Post
1506
Apple released AIMv2 🍏 a family of state-of-the-art open-set vision encoders
apple/aimv2-6720fe1558d94c7805f7688c
> like CLIP, but add a decoder and train on autoregression 🤯
> 19 open models come in 300M, 600M, 1.2B, 2.7B with resolutions of 224, 336, 448
> Load and use with 🤗 transformers
apple/aimv2-6720fe1558d94c7805f7688c
> like CLIP, but add a decoder and train on autoregression 🤯
> 19 open models come in 300M, 600M, 1.2B, 2.7B with resolutions of 224, 336, 448
> Load and use with 🤗 transformers
Post
3134
your hugging face profile now has your recent activities 🤗
Post
1818
🤗 transformers pipelines now support vision language models for easy local inference 🫰🏻
h/t @yonigozlan for shipping this 🎩👏
you can also use inference API to infer hosted vision LMs (via Python, JS and cURL) https://huggingface.co/docs/api-inference/en/tasks/image-text-to-text
h/t @yonigozlan for shipping this 🎩👏
you can also use inference API to infer hosted vision LMs (via Python, JS and cURL) https://huggingface.co/docs/api-inference/en/tasks/image-text-to-text
Post
5034
OmniVision-968M: a new local VLM for edge devices, fast & small but performant
💨 a new vision language model with 9x less image tokens, super efficient
📖 aligned with DPO for reducing hallucinations
⚡️ Apache 2.0 license 🔥
Demo hf.co/spaces/NexaAIDev/omnivlm-dpo-demo
Model https://huggingface.co/NexaAIDev/omnivision-968M
💨 a new vision language model with 9x less image tokens, super efficient
📖 aligned with DPO for reducing hallucinations
⚡️ Apache 2.0 license 🔥
Demo hf.co/spaces/NexaAIDev/omnivlm-dpo-demo
Model https://huggingface.co/NexaAIDev/omnivision-968M
- 4 replies
Post
1971
Amazing past days at open ML, it's raining coding models, let's have a recap 🌧️ Find all models and datasets here
merve/nov-15-releases-67372d0ebdc354756a52ecd0
Models
💻 Coding: Qwen team released two Qwen2.5-Coder checkpoints of 32B and 7B. Infly released OpenCoder: 1.5B and 8B coding models with instruction SFT'd versions and their datasets! 💗
🖼️ Image/Video Gen: Alibaba vision lab released In-context LoRA -- 10 LoRA models on different themes based on Flux. Also Mochi the sota video generation model with A2.0 license now comes natively supported in diffusers 👏
🖼️ VLMs/Multimodal: NexaAIDev released Omnivision 968M a new vision language model aligned with DPO for reducing hallucinations, also comes with GGUF ckpts 👏 Microsoft released LLM2CLIP, a new CLIP-like model with longer context window allowing complex text inputs and better search
🎮 AGI?: Etched released Oasis 500M, a diffusion based open world model that takes keyboard input and outputs gameplay 🤯
Datasets
Common Corpus: A text dataset with 2T tokens with permissive license for EN/FR on various sources: code, science, finance, culture 📖
Models
💻 Coding: Qwen team released two Qwen2.5-Coder checkpoints of 32B and 7B. Infly released OpenCoder: 1.5B and 8B coding models with instruction SFT'd versions and their datasets! 💗
🖼️ Image/Video Gen: Alibaba vision lab released In-context LoRA -- 10 LoRA models on different themes based on Flux. Also Mochi the sota video generation model with A2.0 license now comes natively supported in diffusers 👏
🖼️ VLMs/Multimodal: NexaAIDev released Omnivision 968M a new vision language model aligned with DPO for reducing hallucinations, also comes with GGUF ckpts 👏 Microsoft released LLM2CLIP, a new CLIP-like model with longer context window allowing complex text inputs and better search
🎮 AGI?: Etched released Oasis 500M, a diffusion based open world model that takes keyboard input and outputs gameplay 🤯
Datasets
Common Corpus: A text dataset with 2T tokens with permissive license for EN/FR on various sources: code, science, finance, culture 📖
Post
1671
Microsoft released LLM2CLIP: a CLIP model with longer context window for complex text inputs 🤯
All models with Apache 2.0 license here microsoft/llm2clip-672323a266173cfa40b32d4c
TLDR; they replaced CLIP's text encoder with various LLMs fine-tuned on captioning, better top-k accuracy on retrieval.
This will enable better image-text retrieval, better zero-shot image classification, better vision language models 🔥
Read the paper to learn more: LLM2CLIP: Powerful Language Model Unlock Richer Visual Representation (2411.04997)
All models with Apache 2.0 license here microsoft/llm2clip-672323a266173cfa40b32d4c
TLDR; they replaced CLIP's text encoder with various LLMs fine-tuned on captioning, better top-k accuracy on retrieval.
This will enable better image-text retrieval, better zero-shot image classification, better vision language models 🔥
Read the paper to learn more: LLM2CLIP: Powerful Language Model Unlock Richer Visual Representation (2411.04997)
Post
5431
Another great week in open ML!
Here's a small recap 🫰🏻
Model releases
⏯️ Video Language Models
AI at Meta released Vision-CAIR/LongVU_Qwen2_7B, a new state-of-the-art long video LM model based on DINOv2, SigLIP, Qwen2 and Llama 3.2
💬 Small language models
Hugging Face released HuggingFaceTB/SmolLM2-1.7B, a family of new smol language models with Apache 2.0 license that come in sizes 135M, 360M and 1.7B, along with datasets.
Meta released facebook/MobileLLM-1B, a new family of on-device LLMs of sizes 125M, 350M and 600M
🖼️ Image Generation
Stability AI released stabilityai/stable-diffusion-3.5-medium, a 2B model with commercially permissive license
🖼️💬Any-to-Any
gpt-omni/mini-omni2 is closest reproduction to GPT-4o, a new LLM that can take image-text-audio input and output speech is released!
Dataset releases
🖼️ Spawning/PD12M, a new captioning dataset of 12.4 million examples generated using Florence-2
Here's a small recap 🫰🏻
Model releases
⏯️ Video Language Models
AI at Meta released Vision-CAIR/LongVU_Qwen2_7B, a new state-of-the-art long video LM model based on DINOv2, SigLIP, Qwen2 and Llama 3.2
💬 Small language models
Hugging Face released HuggingFaceTB/SmolLM2-1.7B, a family of new smol language models with Apache 2.0 license that come in sizes 135M, 360M and 1.7B, along with datasets.
Meta released facebook/MobileLLM-1B, a new family of on-device LLMs of sizes 125M, 350M and 600M
🖼️ Image Generation
Stability AI released stabilityai/stable-diffusion-3.5-medium, a 2B model with commercially permissive license
🖼️💬Any-to-Any
gpt-omni/mini-omni2 is closest reproduction to GPT-4o, a new LLM that can take image-text-audio input and output speech is released!
Dataset releases
🖼️ Spawning/PD12M, a new captioning dataset of 12.4 million examples generated using Florence-2