Running
🚀
Top Contributors: Space Likes
Creators of Spaces with the most cumulative all-time likes
Top Contributors: Space Likes
community
AI & ML interests
🚀 Creators of Spaces with the most cumulative all-time likes at the end of each month (users only, no orgs)
Recent Activity
View all activity
TopContributors-SpaceLikes's activity
mvaloatto
updated
a
Space
about 10 hours ago
Post
3900
Google drops Gemini 2.0 Flash Thinking
a new experimental model that unlocks stronger reasoning capabilities and shows its thoughts. The model plans (with thoughts visible), can solve complex problems with Flash speeds, and more
now available in anychat, try it out: akhaliq/anychat
a new experimental model that unlocks stronger reasoning capabilities and shows its thoughts. The model plans (with thoughts visible), can solve complex problems with Flash speeds, and more
now available in anychat, try it out: akhaliq/anychat
jbilcke-hf
posted
an
update
19 days ago
Post
3414
Doing some testing with HunyuanVideo on the Hugging Face Inference Endpoints 🤗
prompt: "a Shiba Inu is acting as a DJ, he wears sunglasses and is mixing and scratching with vinyl discs at a Ibiza sunny sand beach party"
1280x720, 22 steps, 121 frames
There are still some things to iron out regarding speed and memory usage, right now it takes 20min on an A100 (see attached charts)
but you can check it out here:
https://huggingface.co/jbilcke-hf/HunyuanVideo-for-InferenceEndpoints
There are various things I want to try like the 100% diffusers version and other models (LTX-Video..)
prompt: "a Shiba Inu is acting as a DJ, he wears sunglasses and is mixing and scratching with vinyl discs at a Ibiza sunny sand beach party"
1280x720, 22 steps, 121 frames
There are still some things to iron out regarding speed and memory usage, right now it takes 20min on an A100 (see attached charts)
but you can check it out here:
https://huggingface.co/jbilcke-hf/HunyuanVideo-for-InferenceEndpoints
There are various things I want to try like the 100% diffusers version and other models (LTX-Video..)
Post
5114
QwQ-32B-Preview is now available in anychat
A reasoning model that is competitive with OpenAI o1-mini and o1-preview
try it out: akhaliq/anychat
A reasoning model that is competitive with OpenAI o1-mini and o1-preview
try it out: akhaliq/anychat
- 1 reply
Post
3752
Post
2738
anychat
supports chatgpt, gemini, perplexity, claude, meta llama, grok all in one app
try it out there: akhaliq/anychat
supports chatgpt, gemini, perplexity, claude, meta llama, grok all in one app
try it out there: akhaliq/anychat
Post
9759
DimensionX is out for you to try and duplicate 🤗
—> fffiloni/DimensionX
Discuss Paper: DimensionX: Create Any 3D and 4D Scenes from a Single Image with Controllable Video Diffusion (2411.04928)
Examples by the amazing William Lamkin @phanes
—> fffiloni/DimensionX
Discuss Paper: DimensionX: Create Any 3D and 4D Scenes from a Single Image with Controllable Video Diffusion (2411.04928)
Examples by the amazing William Lamkin @phanes
- 4 replies
Post
16479
Visionary Walter Murch (editor for Francis Ford Coppola), in 1999:
“ So let's suppose a technical apotheosis some time in the middle of the 21st century, when it somehow becomes possible for one person to make an entire feature film, with virtual actors. Would this be a good thing?
If the history of oil painting is any guide, the broadest answer would be yes, with the obvious caution to keep a wary eye on the destabilizing effect of following too intently a hermetically personal vision. One need only look at the unraveling of painting or classical music in the 20th century to see the risks.
Let's go even further, and force the issue to its ultimate conclusion by supposing the diabolical invention of a black box that could directly convert a single person's thoughts into a viewable cinematic reality. You would attach a series of electrodes to various points on your skull and simply think the film into existence.
And since we are time-traveling, let us present this hypothetical invention as a Faustian bargain to the future filmmakers of the 21st century. If this box were offered by some mysterious cloaked figure in exchange for your eternal soul, would you take it?
The kind of filmmakers who would accept, even leap, at the offer are driven by the desire to see their own vision on screen in as pure a form as possible. They accept present levels of collaboration as the evil necessary to achieve this vision. Alfred Hitchcock, I imagine, would be one of them, judging from his description of the creative process: "The film is already made in my head before we start shooting."”
—
Read "A Digital Cinema of the Mind? Could Be" by Walter Murch: https://archive.nytimes.com/www.nytimes.com/library/film/050299future-film.html
“ So let's suppose a technical apotheosis some time in the middle of the 21st century, when it somehow becomes possible for one person to make an entire feature film, with virtual actors. Would this be a good thing?
If the history of oil painting is any guide, the broadest answer would be yes, with the obvious caution to keep a wary eye on the destabilizing effect of following too intently a hermetically personal vision. One need only look at the unraveling of painting or classical music in the 20th century to see the risks.
Let's go even further, and force the issue to its ultimate conclusion by supposing the diabolical invention of a black box that could directly convert a single person's thoughts into a viewable cinematic reality. You would attach a series of electrodes to various points on your skull and simply think the film into existence.
And since we are time-traveling, let us present this hypothetical invention as a Faustian bargain to the future filmmakers of the 21st century. If this box were offered by some mysterious cloaked figure in exchange for your eternal soul, would you take it?
The kind of filmmakers who would accept, even leap, at the offer are driven by the desire to see their own vision on screen in as pure a form as possible. They accept present levels of collaboration as the evil necessary to achieve this vision. Alfred Hitchcock, I imagine, would be one of them, judging from his description of the creative process: "The film is already made in my head before we start shooting."”
—
Read "A Digital Cinema of the Mind? Could Be" by Walter Murch: https://archive.nytimes.com/www.nytimes.com/library/film/050299future-film.html
- 1 reply
multimodalart
posted
an
update
6 months ago
Post
20812
New feature 🔥
Image models and LoRAs now have little previews 🤏
If you don't know where to start to find them, I invite you to browse cool LoRAs in the profile of some amazing fine-tuners: @artificialguybr , @alvdansen , @DoctorDiffusion , @e-n-v-y , @KappaNeuro @ostris
Image models and LoRAs now have little previews 🤏
If you don't know where to start to find them, I invite you to browse cool LoRAs in the profile of some amazing fine-tuners: @artificialguybr , @alvdansen , @DoctorDiffusion , @e-n-v-y , @KappaNeuro @ostris
- 2 replies
jbilcke-hf
posted
an
update
6 months ago
Post
29663
I am working to add automatic speech bubbles to the AI Comic Factory!
This is not finished yet - there are a few things to tweak - but so far results are pretty promising!
(it's available through a "beta" toggle)
This is not finished yet - there are a few things to tweak - but so far results are pretty promising!
(it's available through a "beta" toggle)
- 9 replies
Post
20623
Phased Consistency Model
Phased Consistency Model (2405.18407)
The consistency model (CM) has recently made significant progress in accelerating the generation of diffusion models. However, its application to high-resolution, text-conditioned image generation in the latent space (a.k.a., LCM) remains unsatisfactory. In this paper, we identify three key flaws in the current design of LCM. We investigate the reasons behind these limitations and propose the Phased Consistency Model (PCM), which generalizes the design space and addresses all identified limitations. Our evaluations demonstrate that PCM significantly outperforms LCM across 1--16 step generation settings. While PCM is specifically designed for multi-step refinement, it achieves even superior or comparable 1-step generation results to previously state-of-the-art specifically designed 1-step methods. Furthermore, we show that PCM's methodology is versatile and applicable to video generation, enabling us to train the state-of-the-art few-step text-to-video generator.
Phased Consistency Model (2405.18407)
The consistency model (CM) has recently made significant progress in accelerating the generation of diffusion models. However, its application to high-resolution, text-conditioned image generation in the latent space (a.k.a., LCM) remains unsatisfactory. In this paper, we identify three key flaws in the current design of LCM. We investigate the reasons behind these limitations and propose the Phased Consistency Model (PCM), which generalizes the design space and addresses all identified limitations. Our evaluations demonstrate that PCM significantly outperforms LCM across 1--16 step generation settings. While PCM is specifically designed for multi-step refinement, it achieves even superior or comparable 1-step generation results to previously state-of-the-art specifically designed 1-step methods. Furthermore, we show that PCM's methodology is versatile and applicable to video generation, enabling us to train the state-of-the-art few-step text-to-video generator.
Post
19519
🇫🇷
Quel impact de l’IA sur les filières du cinéma, de l’audiovisuel et du jeu vidéo?
Etude prospective à destination des professionnels
— CNC & BearingPoint | 09/04/2024
Si l’Intelligence Artificielle (IA) est utilisée de longue date dans les secteurs du cinéma, de l’audiovisuel et du jeu vidéo, les nouvelles applications de l’IA générative bousculent notre vision de ce dont est capable une machine et possèdent un potentiel de transformation inédit. Elles impressionnent par la qualité de leurs productions et suscitent par conséquent de nombreux débats, entre attentes et appréhensions.
Le CNC a donc décider de lancer un nouvel Observatoire de l’IA Afin de mieux comprendre les usages de l’IA et ses impacts réels sur la filière de l’image. Dans le cadre de cet Observatoire, le CNC a souhaité dresser un premier état des lieux à travers la cartographie des usages actuels ou potentiels de l’IA à chaque étape du processus de création et de diffusion d’une œuvre, en identifiant les opportunités et risques associés, notamment en termes de métiers et d’emploi. Cette étude CNC / Bearing Point en a présenté les principaux enseignements le 6 mars, lors de la journée CNC « Créer, produire, diffuser à l’heure de l’intelligence artificielle ».
Le CNC publie la version augmentée de la cartographie des usages de l’IA dans les filières du cinéma, de l’audiovisuel et du jeu vidéo.
Lien vers la cartographie complète: https://www.cnc.fr/documents/36995/2097582/Cartographie+des+usages+IA_rapport+complet.pdf/96532829-747e-b85e-c74b-af313072cab7?t=1712309387891
Quel impact de l’IA sur les filières du cinéma, de l’audiovisuel et du jeu vidéo?
Etude prospective à destination des professionnels
— CNC & BearingPoint | 09/04/2024
Si l’Intelligence Artificielle (IA) est utilisée de longue date dans les secteurs du cinéma, de l’audiovisuel et du jeu vidéo, les nouvelles applications de l’IA générative bousculent notre vision de ce dont est capable une machine et possèdent un potentiel de transformation inédit. Elles impressionnent par la qualité de leurs productions et suscitent par conséquent de nombreux débats, entre attentes et appréhensions.
Le CNC a donc décider de lancer un nouvel Observatoire de l’IA Afin de mieux comprendre les usages de l’IA et ses impacts réels sur la filière de l’image. Dans le cadre de cet Observatoire, le CNC a souhaité dresser un premier état des lieux à travers la cartographie des usages actuels ou potentiels de l’IA à chaque étape du processus de création et de diffusion d’une œuvre, en identifiant les opportunités et risques associés, notamment en termes de métiers et d’emploi. Cette étude CNC / Bearing Point en a présenté les principaux enseignements le 6 mars, lors de la journée CNC « Créer, produire, diffuser à l’heure de l’intelligence artificielle ».
Le CNC publie la version augmentée de la cartographie des usages de l’IA dans les filières du cinéma, de l’audiovisuel et du jeu vidéo.
Lien vers la cartographie complète: https://www.cnc.fr/documents/36995/2097582/Cartographie+des+usages+IA_rapport+complet.pdf/96532829-747e-b85e-c74b-af313072cab7?t=1712309387891
- 4 replies
Post
20918
Chameleon
Mixed-Modal Early-Fusion Foundation Models
Chameleon: Mixed-Modal Early-Fusion Foundation Models (2405.09818)
We present Chameleon, a family of early-fusion token-based mixed-modal models capable of understanding and generating images and text in any arbitrary sequence. We outline a stable training approach from inception, an alignment recipe, and an architectural parameterization tailored for the early-fusion, token-based, mixed-modal setting. The models are evaluated on a comprehensive range of tasks, including visual question answering, image captioning, text generation, image generation, and long-form mixed modal generation. Chameleon demonstrates broad and general capabilities, including state-of-the-art performance in image captioning tasks, outperforms Llama-2 in text-only tasks while being competitive with models such as Mixtral 8x7B and Gemini-Pro, and performs non-trivial image generation, all in a single model. It also matches or exceeds the performance of much larger models, including Gemini Pro and GPT-4V, according to human judgments on a new long-form mixed-modal generation evaluation, where either the prompt or outputs contain mixed sequences of both images and text. Chameleon marks a significant step forward in a unified modeling of full multimodal documents.
Mixed-Modal Early-Fusion Foundation Models
Chameleon: Mixed-Modal Early-Fusion Foundation Models (2405.09818)
We present Chameleon, a family of early-fusion token-based mixed-modal models capable of understanding and generating images and text in any arbitrary sequence. We outline a stable training approach from inception, an alignment recipe, and an architectural parameterization tailored for the early-fusion, token-based, mixed-modal setting. The models are evaluated on a comprehensive range of tasks, including visual question answering, image captioning, text generation, image generation, and long-form mixed modal generation. Chameleon demonstrates broad and general capabilities, including state-of-the-art performance in image captioning tasks, outperforms Llama-2 in text-only tasks while being competitive with models such as Mixtral 8x7B and Gemini-Pro, and performs non-trivial image generation, all in a single model. It also matches or exceeds the performance of much larger models, including Gemini Pro and GPT-4V, according to human judgments on a new long-form mixed-modal generation evaluation, where either the prompt or outputs contain mixed sequences of both images and text. Chameleon marks a significant step forward in a unified modeling of full multimodal documents.
multimodalart
posted
an
update
8 months ago
Post
26165
The first open Stable Diffusion 3-like architecture model is JUST out 💣 - but it is not SD3! 🤔
It is Tencent-Hunyuan/HunyuanDiT by Tencent, a 1.5B parameter DiT (diffusion transformer) text-to-image model 🖼️✨, trained with multi-lingual CLIP + multi-lingual T5 text-encoders for english 🤝 chinese understanding
Try it out by yourself here ▶️ https://huggingface.co/spaces/multimodalart/HunyuanDiT
(a bit too slow as the model is chunky and the research code isn't super optimized for inference speed yet)
In the paper they claim to be SOTA open source based on human preference evaluation!
It is Tencent-Hunyuan/HunyuanDiT by Tencent, a 1.5B parameter DiT (diffusion transformer) text-to-image model 🖼️✨, trained with multi-lingual CLIP + multi-lingual T5 text-encoders for english 🤝 chinese understanding
Try it out by yourself here ▶️ https://huggingface.co/spaces/multimodalart/HunyuanDiT
(a bit too slow as the model is chunky and the research code isn't super optimized for inference speed yet)
In the paper they claim to be SOTA open source based on human preference evaluation!
jbilcke-hf
posted
an
update
8 months ago
Post
29951
- 11 replies
jbilcke-hf
posted
an
update
8 months ago
Post
7019
If you haven't done it yet,
try my new AI Stories Factory space it is pretty fun!
jbilcke-hf/ai-stories-factory
You can even prompt the visual and music style 💅
try my new AI Stories Factory space it is pretty fun!
jbilcke-hf/ai-stories-factory
You can even prompt the visual and music style 💅
- 2 replies
Post
6289
A Careful Examination of Large Language Model Performance on Grade School Arithmetic
A Careful Examination of Large Language Model Performance on Grade School Arithmetic (2405.00332)
Large language models (LLMs) have achieved impressive success on many benchmarks for mathematical reasoning. However, there is growing concern that some of this performance actually reflects dataset contamination, where data closely resembling benchmark questions leaks into the training data, instead of true reasoning ability. To investigate this claim rigorously, we commission Grade School Math 1000 (GSM1k). GSM1k is designed to mirror the style and complexity of the established GSM8k benchmark, the gold standard for measuring elementary mathematical reasoning. We ensure that the two benchmarks are comparable across important metrics such as human solve rates, number of steps in solution, answer magnitude, and more. When evaluating leading open- and closed-source LLMs on GSM1k, we observe accuracy drops of up to 13%, with several families of models (e.g., Phi and Mistral) showing evidence of systematic overfitting across almost all model sizes. At the same time, many models, especially those on the frontier, (e.g., Gemini/GPT/Claude) show minimal signs of overfitting. Further analysis suggests a positive relationship (Spearman's r^2=0.32) between a model's probability of generating an example from GSM8k and its performance gap between GSM8k and GSM1k, suggesting that many models may have partially memorized GSM8k.
A Careful Examination of Large Language Model Performance on Grade School Arithmetic (2405.00332)
Large language models (LLMs) have achieved impressive success on many benchmarks for mathematical reasoning. However, there is growing concern that some of this performance actually reflects dataset contamination, where data closely resembling benchmark questions leaks into the training data, instead of true reasoning ability. To investigate this claim rigorously, we commission Grade School Math 1000 (GSM1k). GSM1k is designed to mirror the style and complexity of the established GSM8k benchmark, the gold standard for measuring elementary mathematical reasoning. We ensure that the two benchmarks are comparable across important metrics such as human solve rates, number of steps in solution, answer magnitude, and more. When evaluating leading open- and closed-source LLMs on GSM1k, we observe accuracy drops of up to 13%, with several families of models (e.g., Phi and Mistral) showing evidence of systematic overfitting across almost all model sizes. At the same time, many models, especially those on the frontier, (e.g., Gemini/GPT/Claude) show minimal signs of overfitting. Further analysis suggests a positive relationship (Spearman's r^2=0.32) between a model's probability of generating an example from GSM8k and its performance gap between GSM8k and GSM1k, suggesting that many models may have partially memorized GSM8k.
Post
4778
Octopus v4
Graph of language models
Octopus v4: Graph of language models (2404.19296)
Language models have been effective in a wide range of applications, yet the most sophisticated models are often proprietary. For example, GPT-4 by OpenAI and various models by Anthropic are expensive and consume substantial energy. In contrast, the open-source community has produced competitive models, like Llama3. Furthermore, niche-specific smaller language models, such as those tailored for legal, medical or financial tasks, have outperformed their proprietary counterparts. This paper introduces a novel approach that employs functional tokens to integrate multiple open-source models, each optimized for particular tasks. Our newly developed Octopus v4 model leverages functional tokens to intelligently direct user queries to the most appropriate vertical model and reformat the query to achieve the best performance. Octopus v4, an evolution of the Octopus v1, v2, and v3 models, excels in selection and parameter understanding and reformatting. Additionally, we explore the use of graph as a versatile data structure that effectively coordinates multiple open-source models by harnessing the capabilities of the Octopus model and functional tokens.
Graph of language models
Octopus v4: Graph of language models (2404.19296)
Language models have been effective in a wide range of applications, yet the most sophisticated models are often proprietary. For example, GPT-4 by OpenAI and various models by Anthropic are expensive and consume substantial energy. In contrast, the open-source community has produced competitive models, like Llama3. Furthermore, niche-specific smaller language models, such as those tailored for legal, medical or financial tasks, have outperformed their proprietary counterparts. This paper introduces a novel approach that employs functional tokens to integrate multiple open-source models, each optimized for particular tasks. Our newly developed Octopus v4 model leverages functional tokens to intelligently direct user queries to the most appropriate vertical model and reformat the query to achieve the best performance. Octopus v4, an evolution of the Octopus v1, v2, and v3 models, excels in selection and parameter understanding and reformatting. Additionally, we explore the use of graph as a versatile data structure that effectively coordinates multiple open-source models by harnessing the capabilities of the Octopus model and functional tokens.
Post
4666
Layer Skip
Enabling Early Exit Inference and Self-Speculative Decoding
LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding (2404.16710)
We present LayerSkip, an end-to-end solution to speed-up inference of large language models (LLMs). First, during training we apply layer dropout, with low dropout rates for earlier layers and higher dropout rates for later layers, and an early exit loss where all transformer layers share the same exit. Second, during inference, we show that this training recipe increases the accuracy of early exit at earlier layers, without adding any auxiliary layers or modules to the model. Third, we present a novel self-speculative decoding solution where we exit at early layers and verify and correct with remaining layers of the model. Our proposed self-speculative decoding approach has less memory footprint than other speculative decoding approaches and benefits from shared compute and activations of the draft and verification stages. We run experiments on different Llama model sizes on different types of training: pretraining from scratch, continual pretraining, finetuning on specific data domain, and finetuning on specific task. We implement our inference solution and show speedups of up to 2.16x on summarization for CNN/DM documents, 1.82x on coding, and 2.0x on TOPv2 semantic parsing task.
Enabling Early Exit Inference and Self-Speculative Decoding
LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding (2404.16710)
We present LayerSkip, an end-to-end solution to speed-up inference of large language models (LLMs). First, during training we apply layer dropout, with low dropout rates for earlier layers and higher dropout rates for later layers, and an early exit loss where all transformer layers share the same exit. Second, during inference, we show that this training recipe increases the accuracy of early exit at earlier layers, without adding any auxiliary layers or modules to the model. Third, we present a novel self-speculative decoding solution where we exit at early layers and verify and correct with remaining layers of the model. Our proposed self-speculative decoding approach has less memory footprint than other speculative decoding approaches and benefits from shared compute and activations of the draft and verification stages. We run experiments on different Llama model sizes on different types of training: pretraining from scratch, continual pretraining, finetuning on specific data domain, and finetuning on specific task. We implement our inference solution and show speedups of up to 2.16x on summarization for CNN/DM documents, 1.82x on coding, and 2.0x on TOPv2 semantic parsing task.
Post
3535
CatLIP
CLIP-level Visual Recognition Accuracy with 2.7x Faster Pre-training on Web-scale Image-Text Data
CatLIP: CLIP-level Visual Recognition Accuracy with 2.7x Faster Pre-training on Web-scale Image-Text Data (2404.15653)
Contrastive learning has emerged as a transformative method for learning effective visual representations through the alignment of image and text embeddings. However, pairwise similarity computation in contrastive loss between image and text pairs poses computational challenges. This paper presents a novel weakly supervised pre-training of vision models on web-scale image-text data. The proposed method reframes pre-training on image-text data as a classification task. Consequently, it eliminates the need for pairwise similarity computations in contrastive loss, achieving a remarkable 2.7times acceleration in training speed compared to contrastive learning on web-scale data. Through extensive experiments spanning diverse vision tasks, including detection and segmentation, we demonstrate that the proposed method maintains high representation quality.
CLIP-level Visual Recognition Accuracy with 2.7x Faster Pre-training on Web-scale Image-Text Data
CatLIP: CLIP-level Visual Recognition Accuracy with 2.7x Faster Pre-training on Web-scale Image-Text Data (2404.15653)
Contrastive learning has emerged as a transformative method for learning effective visual representations through the alignment of image and text embeddings. However, pairwise similarity computation in contrastive loss between image and text pairs poses computational challenges. This paper presents a novel weakly supervised pre-training of vision models on web-scale image-text data. The proposed method reframes pre-training on image-text data as a classification task. Consequently, it eliminates the need for pairwise similarity computations in contrastive loss, achieving a remarkable 2.7times acceleration in training speed compared to contrastive learning on web-scale data. Through extensive experiments spanning diverse vision tasks, including detection and segmentation, we demonstrate that the proposed method maintains high representation quality.