SmolVLM: Redefining small and efficient multimodal models
Paper
•
2504.05299
•
Published
•
148
EuroPython 2022
non-profit
AI & ML interests
None defined yet.
Recent Activity
View all activity
EuroPython2022's activity
reach-vb
authored
a
paper
4 days ago
Post
1413
AI Vision & SFT Titans 🌟 Turns PDFs into text, snaps pics, and births AI art.
awacke1/TorchTransformers-Diffusion-CV-SFT
1. OCR a grocery list or train a titan while sipping coffee? ☕
2. Camera Snap 📷: Capture life’s chaos—your cat’s face or that weird receipt. Proof you’re a spy!
3. OCR 🔍: PDFs beg for mercy as GPT-4o extracts text.
4. Image Gen 🎨: Prompt “neon superhero me”
5. PDF 📄: Double-page OCR Single-page sniping
Build Titans 🌱: Train tiny AI models. 💪Characters🧑🎨: Craft quirky heroes.
🎥
awacke1/TorchTransformers-Diffusion-CV-SFT
1. OCR a grocery list or train a titan while sipping coffee? ☕
2. Camera Snap 📷: Capture life’s chaos—your cat’s face or that weird receipt. Proof you’re a spy!
3. OCR 🔍: PDFs beg for mercy as GPT-4o extracts text.
4. Image Gen 🎨: Prompt “neon superhero me”
5. PDF 📄: Double-page OCR Single-page sniping
Build Titans 🌱: Train tiny AI models. 💪Characters🧑🎨: Craft quirky heroes.
🎥
osanseviero
authored
a
paper
14 days ago
Post
2240
I introduce MIT license
ML Model Specialize Fine Tuner app "SFT Tiny Titans" 🚀
Demo video with source.
Download, train, SFT, and test your models, easy as 1-2-3!
URL: awacke1/TorchTransformers-NLP-CV-SFT
ML Model Specialize Fine Tuner app "SFT Tiny Titans" 🚀
Demo video with source.
Download, train, SFT, and test your models, easy as 1-2-3!
URL: awacke1/TorchTransformers-NLP-CV-SFT
- 2 replies
ZhangYuanhan
authored
a
paper
about 1 month ago
taesiri
authored
a
paper
about 1 month ago
Post
2479
🚀 Blast into the future with ZaxxonGalaxian – a thrilling 3D action game where you navigate epic battles through towering 3D cityscapes! Face off against relentless swarm bots, climb the leaderboard, and dominate the skies.
awacke1/ZaxxoGalaxian
taesiri
authored
a
paper
about 2 months ago
reach-vb
authored
a
paper
2 months ago
ZhangYuanhan
authored
a
paper
3 months ago
Post
3199
Deep Research Evaluator was asked:
" design a coral defense mechanism that upon sensing say an acid that's causing coral reefs to have a carbon dioxide issue it develops... please create a plan and a design for this\n
"
It picks these three as best combined solution.
1. [Reef-insight: A framework for reef habitat mapping with clustering methods via remote sensing]...
2. Phone a friend: [Learning to Communicate and Collaborate in a Competitive Multi-Agent Setup to Clean the Ocean from Macroplastics]...
3. World Solve: [Dependence of Physiochemical Features on Marine Chlorophyll Analysis with Learning Techniques]
To design a system that allows coralows coral reefs to respond to increased acidity levels in their environment, we can create a network of pH sensors and dispersal units that can detect changes in pH levels and release a base solution to neutralize the acid.
1. pH Sensors: The first component of the system would be a network of pH sensors placed strategically throughout the coral reef. These sensors would be small, durable, and able to withstand the harsh conditions of the ocean. They would be placed at various depths and locations within the reef to ensure accurate and comprehensive monitoring of pH levels.
2. Base Dispersal Units: Once the pH sensors detect a decrease in pH levels, they would trigger the base dispersal units to release a base solution into the water. These units would be strategically placed around the reef and would be able to release a controlled amount of base solution to neutralize the acidity in the water.
3. Water Dispersal Mechanism: The base dispersal units would be connected to a water dispersal mechanism that would allow the base solution to be distributed evenly around the reef. This could be achieved through a series of pipes or channels that would distribute the base solution in a controlled and targeted manner.
" design a coral defense mechanism that upon sensing say an acid that's causing coral reefs to have a carbon dioxide issue it develops... please create a plan and a design for this\n
"
It picks these three as best combined solution.
1. [Reef-insight: A framework for reef habitat mapping with clustering methods via remote sensing]...
2. Phone a friend: [Learning to Communicate and Collaborate in a Competitive Multi-Agent Setup to Clean the Ocean from Macroplastics]...
3. World Solve: [Dependence of Physiochemical Features on Marine Chlorophyll Analysis with Learning Techniques]
To design a system that allows coralows coral reefs to respond to increased acidity levels in their environment, we can create a network of pH sensors and dispersal units that can detect changes in pH levels and release a base solution to neutralize the acid.
1. pH Sensors: The first component of the system would be a network of pH sensors placed strategically throughout the coral reef. These sensors would be small, durable, and able to withstand the harsh conditions of the ocean. They would be placed at various depths and locations within the reef to ensure accurate and comprehensive monitoring of pH levels.
2. Base Dispersal Units: Once the pH sensors detect a decrease in pH levels, they would trigger the base dispersal units to release a base solution into the water. These units would be strategically placed around the reef and would be able to release a controlled amount of base solution to neutralize the acidity in the water.
3. Water Dispersal Mechanism: The base dispersal units would be connected to a water dispersal mechanism that would allow the base solution to be distributed evenly around the reef. This could be achieved through a series of pipes or channels that would distribute the base solution in a controlled and targeted manner.
Post
2685
LLMs and LRMs - Logical Reasoning and Chain of Thought.
This is a read-aloud lecture to answer questions of using language reasoning techniques in advanced AGI style chain of thought AI pipelines.
Produced using DeepResearchEvaluator located here: awacke1/DeepResearchEvaluator
Videos:
https://x.com/Aaron_Wacker/status/1874835790087463063
https://www.youtube.com/watch?v=fW_A1hH_7RM
This is a read-aloud lecture to answer questions of using language reasoning techniques in advanced AGI style chain of thought AI pipelines.
Produced using DeepResearchEvaluator located here: awacke1/DeepResearchEvaluator
Videos:
https://x.com/Aaron_Wacker/status/1874835790087463063
https://www.youtube.com/watch?v=fW_A1hH_7RM
- 1 reply
Post
6332
VLMs are going through quite an open revolution AND on-device friendly sizes:
1. Google DeepMind w/ PaliGemma2 - 3B, 10B & 28B: google/paligemma-2-release-67500e1e1dbfdd4dee27ba48
2. OpenGVLabs w/ InternVL 2.5 - 1B, 2B, 4B, 8B, 26B, 38B & 78B: https://huggingface.co/collections/OpenGVLab/internvl-25-673e1019b66e2218f68d7c1c
3. Qwen w/ Qwen 2 VL - 2B, 7B & 72B: Qwen/qwen2-vl-66cee7455501d7126940800d
4. Microsoft w/ FlorenceVL - 3B & 8B: @jiuhai
5. Moondream2 w/ 0.5B: https://huggingface.co/vikhyatk/
What a time to be alive! 🔥
1. Google DeepMind w/ PaliGemma2 - 3B, 10B & 28B: google/paligemma-2-release-67500e1e1dbfdd4dee27ba48
2. OpenGVLabs w/ InternVL 2.5 - 1B, 2B, 4B, 8B, 26B, 38B & 78B: https://huggingface.co/collections/OpenGVLab/internvl-25-673e1019b66e2218f68d7c1c
3. Qwen w/ Qwen 2 VL - 2B, 7B & 72B: Qwen/qwen2-vl-66cee7455501d7126940800d
4. Microsoft w/ FlorenceVL - 3B & 8B: @jiuhai
5. Moondream2 w/ 0.5B: https://huggingface.co/vikhyatk/
What a time to be alive! 🔥
Post
4766
Massive week for Open AI/ ML:
Mistral Pixtral & Instruct Large - ~123B, 128K context, multilingual, json + function calling & open weights
mistralai/Pixtral-Large-Instruct-2411
mistralai/Mistral-Large-Instruct-2411
Allen AI Tülu 70B & 8B - competive with claude 3.5 haiku, beats all major open models like llama 3.1 70B, qwen 2.5 and nemotron
allenai/tulu-3-models-673b8e0dc3512e30e7dc54f5
allenai/tulu-3-datasets-673b8df14442393f7213f372
Llava o1 - vlm capable of spontaneous, systematic reasoning, similar to GPT-o1, 11B model outperforms gemini-1.5-pro, gpt-4o-mini, and llama-3.2-90B-vision
Xkev/Llama-3.2V-11B-cot
Black Forest Labs Flux.1 tools - four new state of the art model checkpoints & 2 adapters for fill, depth, canny & redux, open weights
reach-vb/black-forest-labs-flux1-6743847bde9997dd26609817
Jina AI Jina CLIP v2 - general purpose multilingual and multimodal (text & image) embedding model, 900M params, 512 x 512 resolution, matroyoshka representations (1024 to 64)
jinaai/jina-clip-v2
Apple AIM v2 & CoreML MobileCLIP - large scale vision encoders outperform CLIP and SigLIP. CoreML optimised MobileCLIP models
apple/aimv2-6720fe1558d94c7805f7688c
apple/coreml-mobileclip
A lot more got released like, OpenScholar (https://huggingface.co/collections/OpenScholar/openscholar-v1-67376a89f6a80f448da411a6), smoltalk ( HuggingFaceTB/smoltalk), Hymba ( nvidia/hymba-673c35516c12c4b98b5e845f), Open ASR Leaderboard ( hf-audio/open_asr_leaderboard) and much more..
Can't wait for the next week! 🤗
Mistral Pixtral & Instruct Large - ~123B, 128K context, multilingual, json + function calling & open weights
mistralai/Pixtral-Large-Instruct-2411
mistralai/Mistral-Large-Instruct-2411
Allen AI Tülu 70B & 8B - competive with claude 3.5 haiku, beats all major open models like llama 3.1 70B, qwen 2.5 and nemotron
allenai/tulu-3-models-673b8e0dc3512e30e7dc54f5
allenai/tulu-3-datasets-673b8df14442393f7213f372
Llava o1 - vlm capable of spontaneous, systematic reasoning, similar to GPT-o1, 11B model outperforms gemini-1.5-pro, gpt-4o-mini, and llama-3.2-90B-vision
Xkev/Llama-3.2V-11B-cot
Black Forest Labs Flux.1 tools - four new state of the art model checkpoints & 2 adapters for fill, depth, canny & redux, open weights
reach-vb/black-forest-labs-flux1-6743847bde9997dd26609817
Jina AI Jina CLIP v2 - general purpose multilingual and multimodal (text & image) embedding model, 900M params, 512 x 512 resolution, matroyoshka representations (1024 to 64)
jinaai/jina-clip-v2
Apple AIM v2 & CoreML MobileCLIP - large scale vision encoders outperform CLIP and SigLIP. CoreML optimised MobileCLIP models
apple/aimv2-6720fe1558d94c7805f7688c
apple/coreml-mobileclip
A lot more got released like, OpenScholar (https://huggingface.co/collections/OpenScholar/openscholar-v1-67376a89f6a80f448da411a6), smoltalk ( HuggingFaceTB/smoltalk), Hymba ( nvidia/hymba-673c35516c12c4b98b5e845f), Open ASR Leaderboard ( hf-audio/open_asr_leaderboard) and much more..
Can't wait for the next week! 🤗
Post
4565
What a brilliant week for Open Source AI!
Qwen 2.5 Coder by Alibaba - 0.5B / 1.5B / 3B / 7B / 14B/ 32B (Base + Instruct) Code generation LLMs, with 32B tackling giants like Gemnini 1.5 Pro, Claude Sonnet
Qwen/qwen25-coder-66eaa22e6f99801bf65b0c2f
LLM2CLIP from Microsoft - Leverage LLMs to train ultra-powerful CLIP models! Boosts performance over the previous SOTA by ~17%
microsoft/llm2clip-672323a266173cfa40b32d4c
Athene v2 Chat & Agent by NexusFlow - SoTA general LLM fine-tuned from Qwen 2.5 72B excels at Chat + Function Calling/ JSON/ Agents
Nexusflow/athene-v2-6735b85e505981a794fb02cc
Orca Agent Instruct by Microsoft - 1 million instruct pairs covering text editing, creative writing, coding, reading comprehension, etc - permissively licensed
microsoft/orca-agentinstruct-1M-v1
Ultravox by FixieAI - 70B/ 8B model approaching GPT4o level, pick any LLM, train an adapter with Whisper as Audio Encoder
reach-vb/ultravox-audio-language-model-release-67373b602af0a52b2a88ae71
JanusFlow 1.3 by DeepSeek - Next iteration of their Unified MultiModal LLM Janus with RectifiedFlow
deepseek-ai/JanusFlow-1.3B
Common Corpus by Pleais - 2,003,039,184,047 multilingual, commercially permissive and high quality tokens!
PleIAs/common_corpus
I'm sure I missed a lot, can't wait for the next week!
Put down in comments what I missed! 🤗
Qwen 2.5 Coder by Alibaba - 0.5B / 1.5B / 3B / 7B / 14B/ 32B (Base + Instruct) Code generation LLMs, with 32B tackling giants like Gemnini 1.5 Pro, Claude Sonnet
Qwen/qwen25-coder-66eaa22e6f99801bf65b0c2f
LLM2CLIP from Microsoft - Leverage LLMs to train ultra-powerful CLIP models! Boosts performance over the previous SOTA by ~17%
microsoft/llm2clip-672323a266173cfa40b32d4c
Athene v2 Chat & Agent by NexusFlow - SoTA general LLM fine-tuned from Qwen 2.5 72B excels at Chat + Function Calling/ JSON/ Agents
Nexusflow/athene-v2-6735b85e505981a794fb02cc
Orca Agent Instruct by Microsoft - 1 million instruct pairs covering text editing, creative writing, coding, reading comprehension, etc - permissively licensed
microsoft/orca-agentinstruct-1M-v1
Ultravox by FixieAI - 70B/ 8B model approaching GPT4o level, pick any LLM, train an adapter with Whisper as Audio Encoder
reach-vb/ultravox-audio-language-model-release-67373b602af0a52b2a88ae71
JanusFlow 1.3 by DeepSeek - Next iteration of their Unified MultiModal LLM Janus with RectifiedFlow
deepseek-ai/JanusFlow-1.3B
Common Corpus by Pleais - 2,003,039,184,047 multilingual, commercially permissive and high quality tokens!
PleIAs/common_corpus
I'm sure I missed a lot, can't wait for the next week!
Put down in comments what I missed! 🤗
Post
1016
🕊️Hope🕊️ and ⚖️Justice⚖️ AI
🚲 Stolen bike in Denver FOUND - Sometimes hope & justice DO prevail.
🎬 So I Created an AI+Art+Music tribute:
-🧠 AI App that Evaluates GPT-4o vs Claude:
awacke1/RescuerOfStolenBikes
https://x.com/Aaron_Wacker/status/1857640877986033980?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1857640877986033980%7Ctwgr%5E203a5022b0eb4c41ee8c1dd9f158330216ac5be1%7Ctwcon%5Es1_c10&ref_url=https%3A%2F%2Fpublish.twitter.com%2F%3Furl%3Dhttps%3A%2F%2Ftwitter.com%2FAaron_Wacker%2Fstatus%2F1857640877986033980
#GPT #Claude #Huggingface
@OpenAI
@AnthropicAI
🚲 Stolen bike in Denver FOUND - Sometimes hope & justice DO prevail.
🎬 So I Created an AI+Art+Music tribute:
-🧠 AI App that Evaluates GPT-4o vs Claude:
awacke1/RescuerOfStolenBikes
https://x.com/Aaron_Wacker/status/1857640877986033980?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1857640877986033980%7Ctwgr%5E203a5022b0eb4c41ee8c1dd9f158330216ac5be1%7Ctwcon%5Es1_c10&ref_url=https%3A%2F%2Fpublish.twitter.com%2F%3Furl%3Dhttps%3A%2F%2Ftwitter.com%2FAaron_Wacker%2Fstatus%2F1857640877986033980
<blockquote class="twitter-tweet"><p lang="en" dir="ltr">QT your 🕊️Hope🕊️ and ⚖️Justice⚖️ art🎨<br><br>🚲 Stolen bike in Denver FOUND! <br> - Sometimes hope & justice DO prevail! <br><br>🎬 Created an AI+Art+Music tribute: <br> -🧠 AI App that Evaluates GPT-4o vs Claude: <a href="https://t.co/odrYdaeizZ">https://t.co/odrYdaeizZ</a><br> <a href="https://twitter.com/hashtag/GPT?src=hash&ref_src=twsrc%5Etfw">#GPT</a> <a href="https://twitter.com/hashtag/Claude?src=hash&ref_src=twsrc%5Etfw">#Claude</a> <a href="https://twitter.com/hashtag/Huggingface?src=hash&ref_src=twsrc%5Etfw">#Huggingface</a> <a href="https://twitter.com/OpenAI?ref_src=twsrc%5Etfw">@OpenAI</a> <a href="https://twitter.com/AnthropicAI?ref_src=twsrc%5Etfw">@AnthropicAI</a> <a href="https://t.co/Q9wGNzLm5C">pic.twitter.com/Q9wGNzLm5C</a></p>— Aaron Wacker (@Aaron_Wacker) <a href="https://twitter.com/Aaron_Wacker/status/1857640877986033980?ref_src=twsrc%5Etfw">November 16, 2024</a></blockquote> <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>#GPT #Claude #Huggingface
@OpenAI
@AnthropicAI
Post
1813
Smol TTS models are here! OuteTTS-0.1-350M - Zero shot voice cloning, built on LLaMa architecture, CC-BY license! 🔥
> Pure language modeling approach to TTS
> Zero-shot voice cloning
> LLaMa architecture w/ Audio tokens (WavTokenizer)
> BONUS: Works on-device w/ llama.cpp ⚡
Three-step approach to TTS:
> Audio tokenization using WavTokenizer (75 tok per second)
> CTC forced alignment for word-to-audio token mapping
> Structured prompt creation w/ transcription, duration, audio tokens
The model is extremely impressive for 350M parameters! Kudos to the
OuteAI team on such a brilliant feat - I'd love to see this be applied on larger data and smarter backbones like SmolLM 🤗
Check out the models here: OuteAI/outetts-6728aa71a53a076e4ba4817c
> Pure language modeling approach to TTS
> Zero-shot voice cloning
> LLaMa architecture w/ Audio tokens (WavTokenizer)
> BONUS: Works on-device w/ llama.cpp ⚡
Three-step approach to TTS:
> Audio tokenization using WavTokenizer (75 tok per second)
> CTC forced alignment for word-to-audio token mapping
> Structured prompt creation w/ transcription, duration, audio tokens
The model is extremely impressive for 350M parameters! Kudos to the
OuteAI team on such a brilliant feat - I'd love to see this be applied on larger data and smarter backbones like SmolLM 🤗
Check out the models here: OuteAI/outetts-6728aa71a53a076e4ba4817c
reach-vb
authored
a
paper
5 months ago
Post
3086
Smol models ftw! AMD released AMD OLMo 1B - beats OpenELM, tiny llama on MT Bench, Alpaca Eval - Apache 2.0 licensed 🔥
> Trained with 1.3 trillion (dolma 1.7) tokens on 16 nodes, each with 4 MI250 GPUs
> Three checkpoints:
- AMD OLMo 1B: Pre-trained model
- AMD OLMo 1B SFT: Supervised fine-tuned on Tulu V2, OpenHermes-2.5, WebInstructSub, and Code-Feedback datasets
- AMD OLMo 1B SFT DPO: Aligned with human preferences using Direct Preference Optimization (DPO) on UltraFeedback dataset
Key Insights:
> Pre-trained with less than half the tokens of OLMo-1B
> Post-training steps include two-phase SFT and DPO alignment
> Data for SFT:
- Phase 1: Tulu V2
- Phase 2: OpenHermes-2.5, WebInstructSub, and Code-Feedback
> Model checkpoints on the Hub & Integrated with Transformers ⚡️
Congratulations & kudos to AMD on a brilliant smol model release! 🤗
amd/amd-olmo-6723e7d04a49116d8ec95070
> Trained with 1.3 trillion (dolma 1.7) tokens on 16 nodes, each with 4 MI250 GPUs
> Three checkpoints:
- AMD OLMo 1B: Pre-trained model
- AMD OLMo 1B SFT: Supervised fine-tuned on Tulu V2, OpenHermes-2.5, WebInstructSub, and Code-Feedback datasets
- AMD OLMo 1B SFT DPO: Aligned with human preferences using Direct Preference Optimization (DPO) on UltraFeedback dataset
Key Insights:
> Pre-trained with less than half the tokens of OLMo-1B
> Post-training steps include two-phase SFT and DPO alignment
> Data for SFT:
- Phase 1: Tulu V2
- Phase 2: OpenHermes-2.5, WebInstructSub, and Code-Feedback
> Model checkpoints on the Hub & Integrated with Transformers ⚡️
Congratulations & kudos to AMD on a brilliant smol model release! 🤗
amd/amd-olmo-6723e7d04a49116d8ec95070
Post
1968
Since 2022 I have been trying to understand how to support advancement of the two best python patterns for AI development which are:
1. Streamlit
2. Gradio
The reason I chose them in this order was the fact that the streamlit library had the timing drop on gradio by being available with near perfection about a year or two before training data tap of GPT.
Nowadays its important that if you want current code to be right on generation it requires understanding of consistency in code method names so no manual intervention is required with each try.
With GPT and Claude being my top two for best AI pair programming models, I gravitate towards streamlit since aside from common repeat errors on cache and experimental functions circa 2022 were not solidified.
Its consistency therefore lacks human correction needs. Old dataset error situations are minimal.
Now, I seek to make it consistent on gradio side. Why? Gradio lapped streamlit for blocks paradigm and API for free which are I feel are amazing features which change software engineering forever.
For a few months I thought BigCode would become the new best model due to its training corpus datasets, yet I never felt it got to market as the next best AI coder model.
I am curious on Gradio's future and how. If the two main models (GPT and Claude) pick up the last few years, I could then code with AI without manual intervention. As it stands today Gradio is better if you could get the best coding models to not repeatedly confuse old syntax as current syntax yet we do live in an imperfect world!
Is anyone using an AI pair programming model that rocks with Gradio's latest syntax? I would like to code with a model that knows how to not miss the advancements and syntax changes that gradio has had in the past few years. Trying grok2 as well.
My IDE coding love is HF. Its hands down faster (100x) than other cloud paradigms. Any tips on models best for gradio coding I can use?
--Aaron
1. Streamlit
2. Gradio
The reason I chose them in this order was the fact that the streamlit library had the timing drop on gradio by being available with near perfection about a year or two before training data tap of GPT.
Nowadays its important that if you want current code to be right on generation it requires understanding of consistency in code method names so no manual intervention is required with each try.
With GPT and Claude being my top two for best AI pair programming models, I gravitate towards streamlit since aside from common repeat errors on cache and experimental functions circa 2022 were not solidified.
Its consistency therefore lacks human correction needs. Old dataset error situations are minimal.
Now, I seek to make it consistent on gradio side. Why? Gradio lapped streamlit for blocks paradigm and API for free which are I feel are amazing features which change software engineering forever.
For a few months I thought BigCode would become the new best model due to its training corpus datasets, yet I never felt it got to market as the next best AI coder model.
I am curious on Gradio's future and how. If the two main models (GPT and Claude) pick up the last few years, I could then code with AI without manual intervention. As it stands today Gradio is better if you could get the best coding models to not repeatedly confuse old syntax as current syntax yet we do live in an imperfect world!
Is anyone using an AI pair programming model that rocks with Gradio's latest syntax? I would like to code with a model that knows how to not miss the advancements and syntax changes that gradio has had in the past few years. Trying grok2 as well.
My IDE coding love is HF. Its hands down faster (100x) than other cloud paradigms. Any tips on models best for gradio coding I can use?
--Aaron
-
- 5 replies