Yaron Lipman
@lipmanya
Followers
4K
Following
1K
Media
25
Statuses
322
Research scientist @AIatMeta (FAIR), prev/visiting @WeizmannScience. Interested in generative models and deep learning of irregular/geometric data.ποΈ
Israel
Joined August 2014
A new (and comprehensive) Flow Matching guide and codebase released! Join us tomorrow at 9:30AM @NeurIPSConf for the FM tutorial to hear more... https://t.co/uaDy00wEw6
https://t.co/ceJlUiTuWO
9
108
521
Come do a PhD with me π! Promise of fun science and great coffee β
31
72
742
New work led by @peholderrieth showing how to transform an already trained flow matching model to a stochastic transition/posterior model that can still be sampled via an efficient ODE solver!
New work: βGLASS Flows: Transition Sampling for Alignment of Flow and Diffusion Modelsβ. GLASS generates images by sampling stochastic Markov transitions with ODEs - allowing us to boost text-image alignment for large-scale models at inference time! https://t.co/unsuG3mYer [1/7]
0
6
61
New work: βGLASS Flows: Transition Sampling for Alignment of Flow and Diffusion Modelsβ. GLASS generates images by sampling stochastic Markov transitions with ODEs - allowing us to boost text-image alignment for large-scale models at inference time! https://t.co/unsuG3mYer [1/7]
4
62
253
We release Code World Model (CWM)! π©βπ»ππ A coding LLM designed to advance code generation research through agentic reasoning and world-model-based planning. Super excited about this release and proud of the teamβs work! π See Gab's post for more info π
(π§΅) Today, we release Meta Code World Model (CWM), a 32-billion-parameter dense LLM that enables novel research on improving code generation through agentic reasoning and planning with world models. https://t.co/BJSUCh2vtg
0
12
50
1/ We released CWM, a 32B dense LLM for coding, agentic use, and, more importantly, to further World-Modeling research. To support this research, we release the pre-training, sft and rl model weights, along with inference code and the tech report. See:
(π§΅) Today, we release Meta Code World Model (CWM), a 32-billion-parameter dense LLM that enables novel research on improving code generation through agentic reasoning and planning with world models. https://t.co/BJSUCh2vtg
1
7
39
(π§΅) Today, we release Meta Code World Model (CWM), a 32-billion-parameter dense LLM that enables novel research on improving code generation through agentic reasoning and planning with world models. https://t.co/BJSUCh2vtg
60
314
2K
The blue vertical lines in the animation indicate blocks start/end; the method generates block-by-block.
0
0
0
A new paper showing a simple method for accelerating LLMs with a short fine-tune and **no** architecture changesβ¦.
Excited to share our work Set Block Decoding! A new paradigm combining next-token-prediction and masked (or discrete diffusion) models, allowing parallel decoding without any architectural changes and with exact KV cache. Arguably one of the simplest ways to accelerate LLMs!
1
2
25
Check out our recent work Set Block Decoding! Super simple modeling, combining NTP and discrete diffusion, allowing parallel decoding without any architectural changes and with exact KV cache! Arxiv:
arxiv.org
Autoregressive next token prediction language models offer powerful capabilities but face significant challenges in practical deployment due to the high computational and memory costs of...
Excited to share our work Set Block Decoding! A new paradigm combining next-token-prediction and masked (or discrete diffusion) models, allowing parallel decoding without any architectural changes and with exact KV cache. Arguably one of the simplest ways to accelerate LLMs!
0
3
23
Excited to share our work Set Block Decoding! A new paradigm combining next-token-prediction and masked (or discrete diffusion) models, allowing parallel decoding without any architectural changes and with exact KV cache. Arguably one of the simplest ways to accelerate LLMs!
3
25
114
DTM vs FMπ Lots of interest in how Difference Transition Matching (DTM) connects to Flow Matching (FM). Here is a short animation that illustrates Theorem 1 in our paper: For a very small step size (1/T), DTM converges to an Euler step of FM.
[1/n] New paper alert! π Excited to introduce ππ«ππ§π¬π’ππ’π¨π§ πππππ‘π’π§π (ππ)! We're replacing short-timestep kernels from Flow Matching/Diffusion with... a generative modelπ€―, achieving SOTA text-2-image generation! @urielsinger @itai_gat @lipmanya
2
50
329
If you're curious to dive deeper into Transition Matching (TM)β¨π, a great starting point is understanding the similarities and differences between ππ’ππππ«ππ§ππ ππ«ππ§π¬π’ππ’π¨π§ πππππ‘π’π§π (πππ) and Flow Matching (FM)π‘.
[1/n] New paper alert! π Excited to introduce ππ«ππ§π¬π’ππ’π¨π§ πππππ‘π’π§π (ππ)! We're replacing short-timestep kernels from Flow Matching/Diffusion with... a generative modelπ€―, achieving SOTA text-2-image generation! @urielsinger @itai_gat @lipmanya
2
17
129
[1/n] New paper alert! π Excited to introduce ππ«ππ§π¬π’ππ’π¨π§ πππππ‘π’π§π (ππ)! We're replacing short-timestep kernels from Flow Matching/Diffusion with... a generative modelπ€―, achieving SOTA text-2-image generation! @urielsinger @itai_gat @lipmanya
5
46
290
Difference Transition Matching (DTM) process is so simple to Illustrate, you can calculate it on a whiteboard! At each step: Draw all lines connecting source and target (shaded) β¬οΈ List those intersecting with the current state (yellow) β¬οΈ Sample a line from the list (green)
[1/n] New paper alert! π Excited to introduce ππ«ππ§π¬π’ππ’π¨π§ πππππ‘π’π§π (ππ)! We're replacing short-timestep kernels from Flow Matching/Diffusion with... a generative modelπ€―, achieving SOTA text-2-image generation! @urielsinger @itai_gat @lipmanya
2
17
138
Introducing Transition Matching (TM) β a new generative paradigm that unifies Flow Matching and autoregressive models into one framework, boosting both quality and speed! Thank you for the great collaboration @shaulneta @itai_gat @lipmanya
[1/n] New paper alert! π Excited to introduce ππ«ππ§π¬π’ππ’π¨π§ πππππ‘π’π§π (ππ)! We're replacing short-timestep kernels from Flow Matching/Diffusion with... a generative modelπ€―, achieving SOTA text-2-image generation! @urielsinger @itai_gat @lipmanya
2
4
22
Check out our team's latest work, led by @urielsinger and @shaulneta!
[1/n] New paper alert! π Excited to introduce ππ«ππ§π¬π’ππ’π¨π§ πππππ‘π’π§π (ππ)! We're replacing short-timestep kernels from Flow Matching/Diffusion with... a generative modelπ€―, achieving SOTA text-2-image generation! @urielsinger @itai_gat @lipmanya
0
2
17
**Transition Matching** is a new iterative generative paradigm using Flow Matching or AR models to transition between generation intermediate states, leading to an improved generation quality and speed!
[1/n] New paper alert! π Excited to introduce ππ«ππ§π¬π’ππ’π¨π§ πππππ‘π’π§π (ππ)! We're replacing short-timestep kernels from Flow Matching/Diffusion with... a generative modelπ€―, achieving SOTA text-2-image generation! @urielsinger @itai_gat @lipmanya
0
19
131
