Research | Post Training of CodeLMs, LMs.
Joined August 2019
- Tweets 377
- Following 475
- Followers 274
- Likes 2,626
74 Photos and videos
Pinned Tweet
6 Apr 2024
How to define Diversity in the context of CodeLMs and Programming Languages ?
1. Diversity is positively correlated with Performance in solving a problem.
2. Shortcomings of diversity in small codeLMs.
3. Code Embedding models don't capture semantics.
reshinthadithyan.github.io/b…
1
9
24
4,503
Reshinth retweeted
Jun 2
We're excited to share Stable-Layers!
We train Qwen-Image-Layered further with RL for improved layerization,
using only feedback from a VLM — no paired supervision required!
Paper: arxiv.org/abs/2605.30257
Project Page: stability-ai.github.io/stabl…
12
51
273
18,074
Reshinth retweeted
1 Oct 2025
🎬 Introducing Stable Cinemetrics, to be presented at NeurIPS 2025.
We present the first taxonomy of professional controls to systematically study and control video generative models through the lens of filmmaking.
Interactive webpage with paper link: stable-cinemetrics.github.io…
🧵
1
3
24
4,116
Reshinth retweeted
13 Jun 2025
🙋♂️ Can RL training address model weaknesses without external distillation?
🚀 Please check our latest work on RL for LLM reasoning!
💯 TL;DR: We propose augmenting RL training with synthetic problems targeting model’s reasoning weaknesses.
📊Qwen2.5-32B: 42.9 → SwS-32B: 68.4
7
37
131
12,191
6 Jun 2025
With CodeLMs scaling actually solved models intrinsically learning internal structural syntactical & semantic information.
132
Reshinth retweeted
6 Jun 2025
Open AI gave a talk on writing software through specs today.
I thought it was my little secret, but seems like quite a few smart builders in the space have also found it's a useful approach.
Now that the secrets out joshuapurtell.com/posts/spec…
20
44
728
76,665
Reshinth retweeted
31 Mar 2025
🧵We just released the #1 open-source agent on the SWE-bench Verified leaderboard by assembling the best of Claude Sonnet 3.7 and O1.
Open-source repo here: github.com/augmentcode/augme…
Here's how we achieved 65.4% success rate on the hardest coding benchmark in the industry: 🧠👇
7
61
276
51,383
23 Feb 2025
ML Twitter lately.
22 Feb 2025
Remember folks: if you aren't a subject matter expert, don't know the context, and have nothing valuable to add to a thread, you always have the option of not replying!
139
Reshinth retweeted
5 Dec 2024
How important is the quality, diversity, and complexity (QDC) of synthetic data for LLM performance? What effect does QDC data composition have on self-improvement?
We just released a comprehensive survey discussing these questions (and many more) 🧵
5
32
111
16,909
Reshinth retweeted
5 Dec 2024
As R&D staff @answerdotai, I work a lot on boosting productivity with AI. A common theme that always comes up is the combination of human AI. This combination proved to be powerful in our new project ShellSage, which is an AI terminal buddy that learns and teaches with you. A 🧵
5
39
201
69,139
Reshinth retweeted
19 Nov 2024
New Anthropic research: Adding Error Bars to Evals.
AI model evaluations don’t usually include statistics or uncertainty. We think they should.
Read the blog post here: anthropic.com/research/stati…
50
297
2,085
756,249
Reshinth retweeted
7 Nov 2024
I'm so excited to be working on this new course from @fastdotai ! Education has always been a huge driving factor in my life. It is surreal that I'm getting to do this as part of my job. Really looking forward to working with students again 🤓
7 Nov 2024
Today, we're announcing that @fastdotai is joining @AnswerdotAI, marking a new phase in making AI accessible.
And we're launching a new a new kind of "AI-first" educational experience, "How To Solve It With Code".
answer.ai/posts/2024-11-07-s…
2
1
18
794
Reshinth retweeted
1 Nov 2024
Our team has been working hard to harness the power of AI to make software more secure.✨🔐
Today we are excited to share a major milestone: our AI agent has discovered its first real-world security vulnerability!
googleprojectzero.blogspot.c…
More info 🧵
8
120
496
158,683
Reshinth retweeted
1 Aug 2024
🔥 I am so damn excited to announce the launch of Black Forest Labs. We set ourselves on a mission to advance state-of-the-art, high-quality generative deep learning models for images and video, and make them available to the broadest audience possible.
Today, we release FLUX.1
1 Aug 2024
We are excited to announce the launch of Black Forest Labs. Our mission is to develop and advance state-of-the-art generative deep learning models for media and to push the boundaries of creativity, efficiency and diversity.
87
153
1,156
386,080
Reshinth retweeted
30 Jul 2024
Fue un absoluto placer hablar con @jaimenovoa de @kfund sobre la industria de la inteligencia artificial, sobre cómo funcionan y se crean los chatbots en términos sencillos, el rol de los datos y el presente y potencial futuro del ecosistema de IA español. open.spotify.com/episode/65w…
4
8
33
6,475
Reshinth retweeted
19 Jun 2024
the training set of ARC is already contaminated...
we have human-written python code to most of the ARC's training set on a github, AND in CoT style prompting that breaks the task down, AND in context with the rendered grids:
iprc-dip.github.io/ANPL/
it's been there since 2023
6
5
66
14,054
Reshinth retweeted
17 Jun 2024
Big life update! I'm super excited to announce I have joined the awesome crew at @answerdotai 🤓
10
4
85
34,147
