Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

We just added speculative decoding for Carbon, so you can now generate DNA sequences 2-3x faster with llama.cpp This enables anyone to process huge genomes locally! GGUF weights and quants:

_lewtun's profile picture

Lewis Tunstall

19,947 subscribers

15,991 Aufrufe • vor 2 Monaten •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

We are releasing Carbon: a crazy fast DNA model Carbon is 275x faster than the next best model. So fast you can process the whole human genome on a single GPU in <2 days. Here are the tricks we used: When modelling DNA sequences a lot of the performance comes down to tokenizing the sequences in a smart way. BPE tokenizer struggle because there are no whitespaces and character (called base in DNA) level tokenizers waste a lot of compute on too many tokens. Carbon is built with a unique tokenizer: we split sequences in chunks of 6 bases, but during both training and inference we can work with single base resolution. That's similar to having word tokens but resolving them at the character level. All possible thanks to the DNA tokens unique structure. The architecture combined with the tokenizer makes the model 275x faster than the previous SoTA (Evo2) at this size. We built an interactive demo so you can explore how the model can generate DNA sequences, investigate the structure of genes, predict the effect of mutations, generate and fold proteins and even reconstruct parts of the tree of life.
1:07

We are releasing Carbon: a crazy fast DNA model Carbon is 275x faster than the next best model. So fast you can process the whole human genome on a single GPU in <2 days. Here are the tricks we used: When modelling DNA sequences a lot of the performance comes down to tokenizing the sequences in a smart way. BPE tokenizer struggle because there are no whitespaces and character (called base in DNA) level tokenizers waste a lot of compute on too many tokens. Carbon is built with a unique tokenizer: we split sequences in chunks of 6 bases, but during both training and inference we can work with single base resolution. That's similar to having word tokens but resolving them at the character level. All possible thanks to the DNA tokens unique structure. The architecture combined with the tokenizer makes the model 275x faster than the previous SoTA (Evo2) at this size. We built an interactive demo so you can explore how the model can generate DNA sequences, investigate the structure of genes, predict the effect of mutations, generate and fold proteins and even reconstruct parts of the tree of life.

Leandro von Werra

404,957 Aufrufe • vor 2 Monaten

Initial DeepSeek v4 and v4 Flash support added to anemll-flash-llama.cpp! 🚀 M5 MAX 128GB can run full DeepSeek-V4 with 1.6T params in original FP8/FP4 weights from SSD without requantization! One-click scripts to convert HF safetensors to dense + MoE sidecar (no GGUF needed) Inference and server examples Prefill and decoding benchmarks Branch and docs here: This is WIP and I plan to add cleanup and more testing this week. Feedback welcome!
0:21

Initial DeepSeek v4 and v4 Flash support added to anemll-flash-llama.cpp! 🚀 M5 MAX 128GB can run full DeepSeek-V4 with 1.6T params in original FP8/FP4 weights from SSD without requantization! One-click scripts to convert HF safetensors to dense + MoE sidecar (no GGUF needed) Inference and server examples Prefill and decoding benchmarks Branch and docs here: This is WIP and I plan to add cleanup and more testing this week. Feedback welcome!

Anemll

32,250 Aufrufe • vor 3 Monaten

wow.. now we can TALK with AI to generate images Midjourney V7 added voice mode, we can generate and edit images in seconds by talking with it we've been waiting for this for a long time
0:31

wow.. now we can TALK with AI to generate images Midjourney V7 added voice mode, we can generate and edit images in seconds by talking with it we've been waiting for this for a long time

el.cine

92,074 Aufrufe • vor 1 Jahr

LFM2-VL support with GGUF and llama.cpp 🥳 You can now run these tiny, hyper-efficient VLMs on your watch! We released quantized checkpoints for LFM2-VL-450M and LFM2-VL-1.6B on Hugging Face
0:59

LFM2-VL support with GGUF and llama.cpp 🥳 You can now run these tiny, hyper-efficient VLMs on your watch! We released quantized checkpoints for LFM2-VL-450M and LFM2-VL-1.6B on Hugging Face

Maxime Labonne

19,947 Aufrufe • vor 11 Monaten

Laguna S 2.1 is built to run locally. We heard the ask for a serious open model in the 80–120B range. Laguna S 2.1 is our answer, and we want it to redefine what this size class can do. Weights ship in BF16, FP8, INT4 and NVFP4, with official GGUF and MLX conversions, plus official DFlash draft models for faster inference. Day-one support across llama.cpp, ollama, Valerie Lee Make Up, SGLang, NVIDIA TRT-LLM, Hugging Face transformers and ZML. Watch it run on an NVIDIA AI DGX Spark.
0:39

Laguna S 2.1 is built to run locally. We heard the ask for a serious open model in the 80–120B range. Laguna S 2.1 is our answer, and we want it to redefine what this size class can do. Weights ship in BF16, FP8, INT4 and NVFP4, with official GGUF and MLX conversions, plus official DFlash draft models for faster inference. Day-one support across llama.cpp, ollama, Valerie Lee Make Up, SGLang, NVIDIA TRT-LLM, Hugging Face transformers and ZML. Watch it run on an NVIDIA AI DGX Spark.

Poolside

31,242 Aufrufe • vor 13 Tagen

this is why open-source wins every time.. open weights allow you to invent new model capabilities, train + generate locally (crucial for NDA'd and personal data), push the limits of what's currently possible + lean into a huge community of experts for advice and inspiration 🖤
2:35

this is why open-source wins every time.. open weights allow you to invent new model capabilities, train + generate locally (crucial for NDA'd and personal data), push the limits of what's currently possible + lean into a huge community of experts for advice and inspiration 🖤

Ingi Erlingsson 🪄

16,877 Aufrufe • vor 2 Monaten

WOW! OPENAI DID IT! They released their first TWO open weights models! This is HUGE news for the entire AI industry. It means you can RIGHT NOW run cutting edge AI models locally on your computer Here's all the benefits and how to set up the models locally yourself:
7:40

WOW! OPENAI DID IT! They released their first TWO open weights models! This is HUGE news for the entire AI industry. It means you can RIGHT NOW run cutting edge AI models locally on your computer Here's all the benefits and how to set up the models locally yourself:

Alex Finn

71,875 Aufrufe • vor 1 Jahr

OpenAI just dropped Sora 2 today it can generate videos with sound now, huge improvement on physics, realism, and control.. honestly no one can tell these are AI videos now here's everything you need to know:
0:25

OpenAI just dropped Sora 2 today it can generate videos with sound now, huge improvement on physics, realism, and control.. honestly no one can tell these are AI videos now here's everything you need to know:

el.cine

38,098 Aufrufe • vor 10 Monaten

We just added Origin UI to MagicPath, so now you can build with these beautiful components. Also, here's how easy it is to use them.
0:34

We just added Origin UI to MagicPath, so now you can build with these beautiful components. Also, here's how easy it is to use them.

Pietro Schirano

18,702 Aufrufe • vor 10 Monaten

gemini 3 flash is so crazy.. it can generate code for an interactive 3d car racing game in ~1 min, control speed and direction with hands, no need code, just text prompts its 100% free &amp; a lot faster than 3 pro how to improve this, anyone?
0:51

gemini 3 flash is so crazy.. it can generate code for an interactive 3d car racing game in ~1 min, control speed and direction with hands, no need code, just text prompts its 100% free &amp; a lot faster than 3 pro how to improve this, anyone?

el.cine

114,398 Aufrufe • vor 7 Monaten

New: We added a /game-dev skill to Manus. You can now describe a game and Manus will build it for you. Working gameplay, deployed to a link you can share with anyone. No game development experience needed. Thanks to Alex Ermolov for the open-source work that made this possible!
0:31

New: We added a /game-dev skill to Manus. You can now describe a game and Manus will build it for you. Working gameplay, deployed to a link you can share with anyone. No game development experience needed. Thanks to Alex Ermolov for the open-source work that made this possible!

Manus

49,097 Aufrufe • vor 24 Tagen

1-Click WAN with ComfyUI Locally Wrote some automation scripts to 1-click download the workflows and models to the correct places as explained in this blog post and.... it just works! Thank you ComfyUI. Just published so anyone can instantly get started with WAN on ComfyUI.
1:23

1-Click WAN with ComfyUI Locally Wrote some automation scripts to 1-click download the workflows and models to the correct places as explained in this blog post and.... it just works! Thank you ComfyUI. Just published so anyone can instantly get started with WAN on ComfyUI.

cocktail peanut

23,292 Aufrufe • vor 1 Jahr

Don't know if this will turn out to be anything, but I've been playing with the idea of adding multiplayer to TLOU 2 on PC. Here we can spawn clients and control them (locally only for now, as I'm just testing) #TheLastOfUsPartII
0:36

Don't know if this will turn out to be anything, but I've been playing with the idea of adding multiplayer to TLOU 2 on PC. Here we can spawn clients and control them (locally only for now, as I'm just testing) #TheLastOfUsPartII

Speclizer

102,009 Aufrufe • vor 6 Monaten

Jan v1 is now more reliable when handling searches. It's an open-source alternative to Perplexity Pro, and it now avoids infinite loops with Jan v1 2509. Performance updates: - Gains on reasoning & creativity benchmarks - Small drop on SimpleQA (91.1 -> 90.7), still competitive with Perplexity Pro The updated version is live on Jan Hub. You can run the model in Jan, llama.cpp, or vLLM. Use the model: - Model: - GGUF: Shoutout to the Qwen team for Qwen3-4B Thinking and Georgi Gerganov for llama.cpp.
0:36

Jan v1 is now more reliable when handling searches. It's an open-source alternative to Perplexity Pro, and it now avoids infinite loops with Jan v1 2509. Performance updates: - Gains on reasoning & creativity benchmarks - Small drop on SimpleQA (91.1 -> 90.7), still competitive with Perplexity Pro The updated version is live on Jan Hub. You can run the model in Jan, llama.cpp, or vLLM. Use the model: - Model: - GGUF: Shoutout to the Qwen team for Qwen3-4B Thinking and Georgi Gerganov for llama.cpp.

👋 Jan

17,759 Aufrufe • vor 10 Monaten

It's 2025 folks. We don't have time for this. You can just be hardline and associate with hardliners. This is the Women's groups on the continent. We are literally being replaced. It's needed. You can cook. You just need to be fearless. Everyone supports these things now. You can just say them. You'll be rewarded for doing so.
1:17

It's 2025 folks. We don't have time for this. You can just be hardline and associate with hardliners. This is the Women's groups on the continent. We are literally being replaced. It's needed. You can cook. You just need to be fearless. Everyone supports these things now. You can just say them. You'll be rewarded for doing so.

Zoomer

92,008 Aufrufe • vor 10 Monaten

Our text-to-image got faster, better and photorealistic 😎 You can generate images in real-time, with an infinite scroll, and ready-to-use prompts and styles. The upgraded version is now available for AI partners and premium users.
0:41

Our text-to-image got faster, better and photorealistic 😎 You can generate images in real-time, with an infinite scroll, and ready-to-use prompts and styles. The upgraded version is now available for AI partners and premium users.

Magnific

4,733,366 Aufrufe • vor 2 Jahren

Grok Imagine V0.9 Now you can start making your Anime opening song so easily. You just need to write your song, use Suno V5 and generate the video with Grok Imagine.
1:12

Grok Imagine V0.9 Now you can start making your Anime opening song so easily. You just need to write your song, use Suno V5 and generate the video with Grok Imagine.

Emily

42,165 Aufrufe • vor 9 Monaten

Quick tip for anyone generating videos with Hailuo 2 or Midjourney Video since they don't generate with any sound. You can generate sound effects for free using MMAUDIO via huggingface. I'm sure a lot of you are familiar with it. It's video to audio synthesis and it's FREE! - Go to this link - Upload your video - Type in a description for the sound - Choose a duration - Click on submit Your video now has sound 👍
0:36

Quick tip for anyone generating videos with Hailuo 2 or Midjourney Video since they don't generate with any sound. You can generate sound effects for free using MMAUDIO via huggingface. I'm sure a lot of you are familiar with it. It's video to audio synthesis and it's FREE! - Go to this link - Upload your video - Type in a description for the sound - Choose a duration - Click on submit Your video now has sound 👍

Travis Davids

30,113 Aufrufe • vor 1 Jahr