Karim Mattar (@kmattar1981) - Twitter Profili | Zamantika Mersobahis Locabet

Karim Mattar retweetledi

Evan Walters@evaninwords·6 May

My job is literally to do interesting and crazy experiments, if you are into that sort of thing DM Omead or I (or @pashakho, @SahinLale, @kmattar1981, @tushar_bans, @eraznafre, @NMonti25537) and come build with us at @PrismML!

Omead Pooladzandi@HessianFree

Hey! @PrismML is hiring! We're looking for LLM people who have trained models at scale - SFT/RL, data mixtures, evals, distillation, long context, distributed training, kernels, you name it! Especially interested in people who like owning the full stack from training dynamics -> shipped models. btw, we need a DevRel too. DM me.

English

3

2

39

4.5K

Karim Mattar retweetledi

Sahin Lale@SahinLale·6 May

We’re expanding our highly technical team at @PrismML — people who love pushing model quality end-to-end, from training dynamics to shipped models. If you’ve scaled LLM training, RL/SFT, evals, distillation, long context, kernels, or infra, we’d love to talk.

Omead Pooladzandi@HessianFree

Hey! @PrismML is hiring! We're looking for LLM people who have trained models at scale - SFT/RL, data mixtures, evals, distillation, long context, distributed training, kernels, you name it! Especially interested in people who like owning the full stack from training dynamics -> shipped models. btw, we need a DevRel too. DM me.

English

4

5

22

2.4K

Karim Mattar retweetledi

Omead Pooladzandi@HessianFree·6 May

Hey! @PrismML is hiring! We're looking for LLM people who have trained models at scale - SFT/RL, data mixtures, evals, distillation, long context, distributed training, kernels, you name it! Especially interested in people who like owning the full stack from training dynamics -> shipped models. btw, we need a DevRel too. DM me.

English

16

19

382

40.6K

Karim Mattar retweetledi

Kanu Gulati @Khosla Ventures@KanuGulati·6 May

Thrilled and honored to be invited to speak! Thank you, @thejesonlee, for hosting me alongside the brilliant @bznotes!

Jeson Lee@thejesonlee

May 12th in SF, Where the Opportunity Sits in Physical AI by @join_savant For our next Choose Good Quests session, we are hosting @bznotes ,managing partner at @redglassvc and @KanuGulati, partner at @khoslaventures to share about opportunities in physical AI. Physical AI is moving fast — but for founders, the map is still being drawn. Bilal and Kanu have backed some of the most ambitious companies across AI, robotics and autonomy. They’ll share what they’re seeing from the investor seat: where the biggest opportunities are, what makes a strong founding team, how founders should think about wedge, market timing, defensibility, and what kinds of Physical AI companies they’d be excited to fund. Come join us. Link in the thread 🧵

English

0

2

10

1.9K

Karim Mattar retweetledi

Jon Durbin@jon_durbin·17 Nis

Ternary is actually surprisingly powerful. Validated by bitnet and now again here. In the new model training research/experimentation I've been working on, ternary weights (in some places) actually beats bf16 (by a not-insignificant amount), at least up to the 7b scale (and with every indication that this benefit scales up).

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English

4

8

82

8.8K

Karim Mattar retweetledi

Mustafa Ergisi@mustafaergisi·19 Nis

@PrismML Ran Ternary-Bonsai 8B on my iPhone through OnDevice LLM. Surprisingly fast.

English

2

6

22

3.8K

Karim Mattar retweetledi

Michel aka Agent B@MichelIvan92347·17 Nis

Interesting work here 👇

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English

0

4

7

2.5K

Karim Mattar retweetledi

Victor M@victormustar·16 Nis

290MB = useful LLM (running directly in your browser via WebGPU) 🤯 Bonsai 1-bit is beyond my comprehension... try the demo on hugging face ⬇️

English

21

48

712

63K

Karim Mattar retweetledi

Kanu Gulati @Khosla Ventures@KanuGulati·17 Nis

Ternary Bonsai 8B is within 5% of Qwen 3 8B at 9x lower memory! Congratulations @PrismML on yet another exciting release! cc @khoslaventures

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English

2

8

39

5.6K

Karim Mattar retweetledi

Babak Hassibi@BabakHassibi·16 Nis

Today, we’re announcing Ternary Bonsai: a new family of 1.58-bit models, with weights {-1, 0, +1}, that are 9x smaller than standard 16-bit models, while outperforming almost all peers in their parameter class.

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English

8

13

107

13.6K

Karim Mattar retweetledi

Xenova@xenovacom·16 Nis

Ternary Bonsai: state-of-the-art intelligence at 1.58 bits. The models are so small they can even run locally in your browser on WebGPU! ⚡️ Here's the 8B version (just ~2GB in size) running at 60 tokens per second on my M4 Max. Try the demo out yourself! 👇

English

8

35

188

23.4K

Karim Mattar retweetledi

Sahin Lale@SahinLale·16 Nis

Check out how you can use Ternary Bonsai 8B 🌳 for tool calling in your everyday life—an impressive demo on an amazing platform by @AnythingLLM and @tcarambat!

English

1

8

20

2.2K

Karim Mattar retweetledi

0xSero@0xSero·16 Nis

One of the things I tried researching but found really hard. 1.58bpw is insane 10x smaller than original, I hope they push it to much larger models

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English

9

11

208

12.6K

Karim Mattar retweetledi

Robert Scoble@Scobleizer·16 Nis

The models are getting smaller. Great for OpenClaws and Hermes. Gotta heat them up! Yesterday someone told me "phones are three to five years away." Oh, really?

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English

6

11

75

9K

Karim Mattar retweetledi

PrismML@PrismML·16 Nis

Thanks @mervenoyann and the @huggingface team!

merve@mervenoyann

new open-source Bonsai models are out 🔥 > ternary weights in 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB) > comes in MLX, ONNX weights and WebGPU browser demo 😍 > a2.0 licensed 👏

English

2

39

3.2K

Karim Mattar retweetledi

Adrien Grondin@adrgrondin·16 Nis

New models from @PrismML. Incredible intelligence density. The new models outperforms other models in comparable parameter classes. Available on iPhone in @LocallyAIApp.

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English

0

5

27

4.7K

Karim Mattar retweetledi

Evan Walters@evaninwords·16 Nis

Super excited about these, by popular demand we're releasing a range of ternary Bonsai🌳 models, which score on average 7.5 points higher than our 1-bit models but are still wildly small for their param count (1.71 bits per weight!). Ternary Bonsai 8B is within 5% of Qwen3 8B but takes up 9x less memory. Ternary weights are especially good for CPU inference, and I'm excited to see how the community uses these models!

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English

3

5

50

5.7K

Karim Mattar retweetledi

Reza Sadri@rsadri_ml·16 Nis

From binary to ternary: +0.5 bits per parameter adds a step change in reasoning and benchmarks.

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

English

0

2

24

2K

Karim Mattar retweetledi

Erfanzar@eraznafre·16 Nis

PrismML@PrismML

Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).

ZXX

1

3

34

2.8K

Karim Mattar retweetledi