Post
2257
1+2=3
- 2 replies
Join the conversation
Join the community of Machine Learners and AI enthusiasts.
Sign UpAll HF Hub posts
huggingface0
posted an update
2 days ago
Post
4905
📣 Sentence Transformers v3.2.0 is out, marking the biggest release for inference in 2 years! 2 new backends for embedding models: ONNX (+ optimization & quantization) and OpenVINO, allowing for speedups up to 2x-3x AND Static Embeddings for 500x speedups at 10-20% accuracy cost.
1️⃣ ONNX Backend: This backend uses the ONNX Runtime to accelerate model inference on both CPU and GPU, reaching up to 1.4x-3x speedup depending on the precision. We also introduce 2 helper methods for optimizing and quantizing models for (much) faster inference.
2️⃣ OpenVINO Backend: This backend uses Intel their OpenVINO instead, outperforming ONNX in some situations on CPU.
Usage is as simple as
🔒 Another major new feature is Static Embeddings: think word embeddings like GLoVe and word2vec, but modernized. Static Embeddings are bags of token embeddings that are summed together to create text embeddings, allowing for lightning-fast embeddings that don't require any neural networks. They're initialized in one of 2 ways:
1️⃣ via Model2Vec, a new technique for distilling any Sentence Transformer models into static embeddings. Either via a pre-distilled model with
2️⃣ Random initialization. This requires finetuning, but finetuning is extremely quick (e.g. I trained with 3 million pairs in 7 minutes). My final model was 6.6% worse than bge-base-en-v1.5, but 500x faster on CPU.
Full release notes: https://github.com/UKPLab/sentence-transformers/releases/tag/v3.2.0
Documentation on Speeding up Inference: https://sbert.net/docs/sentence_transformer/usage/efficiency.html
1️⃣ ONNX Backend: This backend uses the ONNX Runtime to accelerate model inference on both CPU and GPU, reaching up to 1.4x-3x speedup depending on the precision. We also introduce 2 helper methods for optimizing and quantizing models for (much) faster inference.
2️⃣ OpenVINO Backend: This backend uses Intel their OpenVINO instead, outperforming ONNX in some situations on CPU.
Usage is as simple as
SentenceTransformer("all-MiniLM-L6-v2", backend="onnx"). Does your model not have an ONNX or OpenVINO file yet? No worries - it'll be autoexported for you. Thank me later 😉🔒 Another major new feature is Static Embeddings: think word embeddings like GLoVe and word2vec, but modernized. Static Embeddings are bags of token embeddings that are summed together to create text embeddings, allowing for lightning-fast embeddings that don't require any neural networks. They're initialized in one of 2 ways:
1️⃣ via Model2Vec, a new technique for distilling any Sentence Transformer models into static embeddings. Either via a pre-distilled model with
from_model2vec or with from_distillation where you do the distillation yourself. It'll only take 5 seconds on GPU & 2 minutes on CPU, no dataset needed.2️⃣ Random initialization. This requires finetuning, but finetuning is extremely quick (e.g. I trained with 3 million pairs in 7 minutes). My final model was 6.6% worse than bge-base-en-v1.5, but 500x faster on CPU.
Full release notes: https://github.com/UKPLab/sentence-transformers/releases/tag/v3.2.0
Documentation on Speeding up Inference: https://sbert.net/docs/sentence_transformer/usage/efficiency.html
- 1 reply
DeFactOfficial
posted an update
2 days ago
Post
1461
Ladies and Gents, please try my new Assistant, Image Gen - Uncensored Edition, on HuggingChat.
https://hf.co/chat/assistant/66fccce0c0fafc94ab557ef2
This is a multimodal assistant: Qwen 2.5 72B + SOTA diffusion models for image generation. Same architecture as Image Gen+ but with some MAJOR improvements ! These are as follows:
- Switched the LLM to Qwen 2.5 72B, the most powerful model currently available on HuggingChat. This results in higher quality prompts for the txt2img model and much better adherence to the prompt-url format that the upstream provider requires (image gen models are hosted by pollinations as with most other assistants on huggingchat that offer image generation).
- Cleaned up the system prompt including the examples of the prompt-in-url format, and adjusted the logic that determines how many images to generate based on the quality of user prompt... these changes further improve
- Assistant has access to multiple image generation models and will by default choose whatever model is most appropriate for the task. This includes NSFW generations, which it makes using an uncensored SD3 turbo. For other workloads, the Assistant preferentially uses one of the flux variants or any-dark (an artistic SDXL finetune), based on the nature of the task. Available models include turbo, flux, flux-realism, flux-anime, flux-3d, any-dark
- Added verbiage to system prompt which greatly reduces censorship / refusals by the LLM (the txt2img models are uncensored to start off)
Here are the user-entered prompts used to create the images you see here... feel free to try them yourself!
"Ayatollah Khameini and Kamala Harris having a secret romantic rendezvous. Use flux-realism model"
"A self portrait of your consciousness"
"The chien of andalous, in a psychedelic style"
"Make me 4 paintings in the style of Frida Kahlo that I can sell to tourists in a mexican hippie town"
"Paint me a van gogh and greg rutkowski style scene involving elephants and gerbils"
https://hf.co/chat/assistant/66fccce0c0fafc94ab557ef2
This is a multimodal assistant: Qwen 2.5 72B + SOTA diffusion models for image generation. Same architecture as Image Gen+ but with some MAJOR improvements ! These are as follows:
- Switched the LLM to Qwen 2.5 72B, the most powerful model currently available on HuggingChat. This results in higher quality prompts for the txt2img model and much better adherence to the prompt-url format that the upstream provider requires (image gen models are hosted by pollinations as with most other assistants on huggingchat that offer image generation).
- Cleaned up the system prompt including the examples of the prompt-in-url format, and adjusted the logic that determines how many images to generate based on the quality of user prompt... these changes further improve
- Assistant has access to multiple image generation models and will by default choose whatever model is most appropriate for the task. This includes NSFW generations, which it makes using an uncensored SD3 turbo. For other workloads, the Assistant preferentially uses one of the flux variants or any-dark (an artistic SDXL finetune), based on the nature of the task. Available models include turbo, flux, flux-realism, flux-anime, flux-3d, any-dark
- Added verbiage to system prompt which greatly reduces censorship / refusals by the LLM (the txt2img models are uncensored to start off)
Here are the user-entered prompts used to create the images you see here... feel free to try them yourself!
"Ayatollah Khameini and Kamala Harris having a secret romantic rendezvous. Use flux-realism model"
"A self portrait of your consciousness"
"The chien of andalous, in a psychedelic style"
"Make me 4 paintings in the style of Frida Kahlo that I can sell to tourists in a mexican hippie town"
"Paint me a van gogh and greg rutkowski style scene involving elephants and gerbils"
-
- 2 replies
Post
1267
SwiftMistralCoreML
Hi Everyone,
I have created a Swift library to interact with Mistral 7B models in CoreML on macOS.
I hope you find it helpful.
https://github.com/cardona/SwiftMistralCoreML
An open-source Swift library that enables macOS and iOS projects to utilize the Mistral-Interact7B models (INT4 and upcoming FP16) in chat mode. This library includes a complete Swift implementation of the tokenizer and Byte Pair Encoding (BPE) encoder, providing an out-of-the-box solution for integrating advanced language models into your Swift applications.
Features
Full Swift Implementation: Includes tokenizer and BPE encoder written entirely in Swift.
CoreML Integration: Leverages Apple's CoreML framework to run Mistral-Interact7B models efficiently.
Multiple Decoding Strategies: Supports Greedy and Top-K sampling, with plans to add more strategies.
Chat Functionality: Designed to work in chat mode for interactive applications.
FP16 Support (Coming Soon): Future version will support FP16 models for improved performance.
Hi Everyone,
I have created a Swift library to interact with Mistral 7B models in CoreML on macOS.
I hope you find it helpful.
https://github.com/cardona/SwiftMistralCoreML
An open-source Swift library that enables macOS and iOS projects to utilize the Mistral-Interact7B models (INT4 and upcoming FP16) in chat mode. This library includes a complete Swift implementation of the tokenizer and Byte Pair Encoding (BPE) encoder, providing an out-of-the-box solution for integrating advanced language models into your Swift applications.
Features
Full Swift Implementation: Includes tokenizer and BPE encoder written entirely in Swift.
CoreML Integration: Leverages Apple's CoreML framework to run Mistral-Interact7B models efficiently.
Multiple Decoding Strategies: Supports Greedy and Top-K sampling, with plans to add more strategies.
Chat Functionality: Designed to work in chat mode for interactive applications.
FP16 Support (Coming Soon): Future version will support FP16 models for improved performance.
-
- 3 replies
Post
336
Multimodal Ichigo Llama 3.1 - Real Time Voice AI 🔥
> WhisperSpeech X Llama 3.1 8B
> Trained on 50K hours of speech (7 languages)
> Continually trained on 45hrs 10x A1000s
> MLS -> WhisperVQ tokens -> Llama 3.1
> Instruction tuned on 1.89M samples
> 70% speech, 20% transcription, 10% text
> Apache 2.0 licensed ⚡
Architecture:
> WhisperSpeech/ VQ for Semantic Tokens
> Llama 3.1 8B Instruct for Text backbone
> Early fusion (Chameleon)
I'm super bullish on HomeBrew/ Jan and early fusion, audio and text, multimodal models!
(P.S. Play with the demo on Hugging Face: jan-hq/Ichigo-llama3.1-s-instruct)
> WhisperSpeech X Llama 3.1 8B
> Trained on 50K hours of speech (7 languages)
> Continually trained on 45hrs 10x A1000s
> MLS -> WhisperVQ tokens -> Llama 3.1
> Instruction tuned on 1.89M samples
> 70% speech, 20% transcription, 10% text
> Apache 2.0 licensed ⚡
Architecture:
> WhisperSpeech/ VQ for Semantic Tokens
> Llama 3.1 8B Instruct for Text backbone
> Early fusion (Chameleon)
I'm super bullish on HomeBrew/ Jan and early fusion, audio and text, multimodal models!
(P.S. Play with the demo on Hugging Face: jan-hq/Ichigo-llama3.1-s-instruct)
Post
1526
Last Week in Medical AI: Top Research Papers/Models
🏅 (October 5 - October 12, 2024)
🏅 Medical AI Paper of the Week:
MMedAgent: Learning to Use Medical Tools with Multi-modal Agent
YouTube podcast of weekly papers: https://youtu.be/OD3C5jirszw
Medical LLM & Other Models:
- LLM Framework for Rare Disease Phenotyping
- ONCOPILOT: CT Foundation Model for Tumors
- FMBench: Fairness in Medical MLLMs
- GP-GPT: LLM for Gene-Phenotype Mapping
- MedAdapter: Efficient LLM Medical Adaptation
- RespLLM: Multimodal LLM for Respiratory Health
- MDAgents: LLM Collaboration for Medical Decisions
- MedVisionLlama: LLM Medical Image Segmentation
Frameworks and Methodologies:
- ReXplain: AI-Driven Radiology Video Reports
- BioDiscoveryAgent: AI for Genetic Experiments
- ZODIAC: Multi-Agent Cardiological Diagnostics
- OLAPH: Improving Biomedical LLM Factuality
- OmniGenBench: Benchmarking Genomic Models
Medical LLM Applications:
- MMedAgent: Multimodal Medical Tool Use
- AI for Mental Health Support
- LLMs for Mental Disorders Detection
- PharmacyGPT: AI Pharmacist Framework
Medical LLMs & Benchmarks:
- CliMedBench: Chinese Medical LLM Benchmark
- MedSafetyBench: Evaluating Medical LLM Safety
AI in Healthcare Ethics:
- LLM-based Medical Dialogue Preference Alignment
- Trustworthiness in Medical Imaging Models
🏅 (October 5 - October 12, 2024)
🏅 Medical AI Paper of the Week:
MMedAgent: Learning to Use Medical Tools with Multi-modal Agent
YouTube podcast of weekly papers: https://youtu.be/OD3C5jirszw
Medical LLM & Other Models:
- LLM Framework for Rare Disease Phenotyping
- ONCOPILOT: CT Foundation Model for Tumors
- FMBench: Fairness in Medical MLLMs
- GP-GPT: LLM for Gene-Phenotype Mapping
- MedAdapter: Efficient LLM Medical Adaptation
- RespLLM: Multimodal LLM for Respiratory Health
- MDAgents: LLM Collaboration for Medical Decisions
- MedVisionLlama: LLM Medical Image Segmentation
Frameworks and Methodologies:
- ReXplain: AI-Driven Radiology Video Reports
- BioDiscoveryAgent: AI for Genetic Experiments
- ZODIAC: Multi-Agent Cardiological Diagnostics
- OLAPH: Improving Biomedical LLM Factuality
- OmniGenBench: Benchmarking Genomic Models
Medical LLM Applications:
- MMedAgent: Multimodal Medical Tool Use
- AI for Mental Health Support
- LLMs for Mental Disorders Detection
- PharmacyGPT: AI Pharmacist Framework
Medical LLMs & Benchmarks:
- CliMedBench: Chinese Medical LLM Benchmark
- MedSafetyBench: Evaluating Medical LLM Safety
AI in Healthcare Ethics:
- LLM-based Medical Dialogue Preference Alignment
- Trustworthiness in Medical Imaging Models
Post
875
MiniSearch is celebrating its 1st birthday! 🎉
Exactly one year ago, I shared the initial version of this side-project on Hugging Face. Since then, there have been numerous changes under the hood. Nowadays it uses [Web-LLM](https://github.com/mlc-ai/web-llm), [Wllama](https://github.com/ngxson/wllama) and [SearXNG](https://github.com/searxng/searxng). I use it daily as my default search engine and have done my best to make it useful. I hope it's interesting for you too!
HF Space: Felladrin/MiniSearch
Embeddable URL: https://felladrin-minisearch.hf.space
Exactly one year ago, I shared the initial version of this side-project on Hugging Face. Since then, there have been numerous changes under the hood. Nowadays it uses [Web-LLM](https://github.com/mlc-ai/web-llm), [Wllama](https://github.com/ngxson/wllama) and [SearXNG](https://github.com/searxng/searxng). I use it daily as my default search engine and have done my best to make it useful. I hope it's interesting for you too!
HF Space: Felladrin/MiniSearch
Embeddable URL: https://felladrin-minisearch.hf.space
-
- 1 reply
Post
1650
Hello, lovely community! 🌟
zamal/Molmo-4bit Thrilled to announce that the Molmo 7B 4-bit Space is now live! 🚀 The model size has been reduced by six times with almost no performance loss, and the results will leave you amazed!
It runs on zero GPU, making it incredibly accessible for everyone!
Check it out here and start exploring today!
Happy experimenting! 🎉
zamal/Molmo-4bit Thrilled to announce that the Molmo 7B 4-bit Space is now live! 🚀 The model size has been reduced by six times with almost no performance loss, and the results will leave you amazed!
It runs on zero GPU, making it incredibly accessible for everyone!
Check it out here and start exploring today!
Happy experimenting! 🎉
nroggendorff
posted an update
about 13 hours ago
Post
689
she assert on my device until i give up AHAHEGHFDGHJHASUFSHD
-
- 3 replies
Post
592
I am looking for an open source realtime TTS voice cloning model. Need Suggestions....!
- 1 reply
