Raghul Chandramouli

Just a Random guy doing AI Research ;p

[LiteFold @try_litefold]

A lab gets better when the last experiment changes the next design.

Using LiteFold, we ran a peptide campaign against BHRF1. We got 47% in hit rate.

That measurement is the whole point of LiteFold: your data, your target, the next round smarter than the last.

Lots of blogs and technical reports are on the way!

[Raghul Chandramouli @Raghulchan2004]

@llawlie2924 mostly pen&paper, and lot of experimentation

[Raghul Chandramouli @Raghulchan2004]

I have been following!! this course!, changed my life for real while doing Deep learning

Stanford cs312, its a must for all

#Deeplearning #Stanford #learning

https://t.co/TW0L8rFV3d

[Raghul Chandramouli @Raghulchan2004]

@CausalFlops28 @Pritam_ai sure! why not

[Raghul Chandramouli @Raghulchan2004]

@777BHAVYA Modal labs!!

[Raghul Chandramouli @Raghulchan2004]

@Pritam_ai That would be amazing

[Raghul Chandramouli @Raghulchan2004]

@Pritam_ai Nice

[Raghul Chandramouli @Raghulchan2004]

@Henryf1w @grok can you answer him!

[Raghul Chandramouli @Raghulchan2004]

@Joshua_WD exactly @Joshua_WD I believe this is the most important aspect! in understanding, it always seems like it always creates this missteps of understanding my intentions

[Raghul Chandramouli @Raghulchan2004]

So have been using Fable 5.2! it's making a lot of chronologically wrong decision with respect to my intent, that even models like OPUS does not do!! its like ASTRA all over again! let's see! if they patch it up

#AI #Anthropic

[Raghul Chandramouli @Raghulchan2004]

@imranye Or into AI safety!! Wherever feels like natural! :)

[Raghul Chandramouli @Raghulchan2004]

@boardyai Cursor

[Raghul Chandramouli @Raghulchan2004]

@169Pi_ai So it’s post trained on qwen 3.5

[Raghul Chandramouli @Raghulchan2004]

So small models are now 500B param!! is this the new norm, Inkling 2 - small!

[Raghul Chandramouli @Raghulchan2004]

Another Successful PR merged in Diffusers

Learnt crap ton! just reading and going through their repo!!! More to come

[Raghul Chandramouli @Raghulchan2004]

https://t.co/2cZWcueBmq

A very good explainer on why! CUDA!

[Raghul Chandramouli @Raghulchan2004]

What the fuck!!!

[Rikiya Takehi @rikiyatakehi]

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.

We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: https://t.co/bOWCcFhCVq

[Raghul Chandramouli @Raghulchan2004]

@TensorTonic Noice!!! Also should read thr ilya’s list

[Raghul Chandramouli @Raghulchan2004]

@mirendil How does this differe from Claude for science