Shubham Toshniwal
@ShubhamToshniw6
Followers
328
Following
1K
Media
2
Statuses
81
Research Scientist @ NVIDIA. ex-Meta, TTIC, IIT Kanpur
NYC
Joined March 2021
RT @jaseweston: 🌉 Bridging Offline & Online RL for LLMs 🌉.📝: New paper shows on verifiable & non-verifiable tasks:….
0
96
0
What a blog! Need more such checks.
Confused about recent LLM RL results where models improve without any ground-truth signal? We were too. Until we looked at the reported numbers of the Pre-RL models and realized they were serverely underreported across papers. We compiled discrepancies in a blog below🧵👇
0
0
1
RT @jaseweston: 🚨Announcing RAM 2 workshop @ COLM25 - call for papers🚨 .- 10 years on, we present the sequel to the classic RAM🐏 (Reasoning….
0
30
0
RT @kuchaev: NeMo RL is now open source! It replaces NeMo-Aligner and is the toolkit we use to post train next generations of our models. G….
0
64
0
RT @HaseoX94: We finally (!) released all our SOTA Code Reasoning models ! Play around with them and get Better scores than QwQ* with 20-30….
0
3
0
RT @reach_vb: NVIDIA just open sourced Open Code Reasoning models - 32B, 14B AND 7B - APACHE 2.0 licensed 🔥. > Beats O3 mini & O1 (low) on….
0
139
0
RT @gonedarragh: 🤩 🤩 AIMO2 keeps improving #aimoprize @igtmn @ShubhamToshniw6 @i_vainn @kagglingdieter.
0
6
0
RT @_weiping: Introducing AceMath-RL-Nemotron-7B, an open math model trained with reinforcement learning from the SFT-only checkpoint: Deep….
0
22
0
RT @gonedarragh: AIMO-2 Winning Solution: Building State-of-the-Art Mathematical Reasoning Models with OpenMathReasoning dataset.abs: https….
0
15
0
RT @kagglingdieter: Happy to announce that we published our 🥇 1st place winning model for the AI Math Olympiad (and smaller/ bigger varian….
0
53
0
RT @gonedarragh: Our team, NemoSkills, is presumptive winner of AIMO2. Outstanding organization from AIMO, Kaggle, XTX markets, @friederrrr….
0
11
0
RT @NVIDIAAIDev: 🎉 Huge congrats to our NVIDIA team “NemoSkills” for winning the AIMO-2 Competition 🏆 on @Kaggle. Their system solved 34….
0
20
0
RT @yoavgo: @srush_nlp or maybe in other words: i feel that with DL, our previous NLP training was helpful and allowed us to identify oppor….
0
1
0
RT @wellecks: Check out our new benchmark for an increasingly important capability: generating synthetic data. Among other insights, it tur….
0
4
0
"A research team led by neurobiologist Margaret Livingstone trained three rhesus macaques to identify symbols representing the numbers zero to 25. They then taught the test subjects how to perform addition. According to the study, all three monkeys were on average capable of.
1
0
6
RT @fredahshi: I’d always be proud of receiving my PhD from TTIC, a magic place which gives you the most unique (in a positive sense, of co….
0
2
0
RT @kuchaev: Llama-3.1-Nemotron-70B-Instruct model aligned by our team is now live on leaderboard with overall rank….
0
20
0
RT @iScienceLuvr: Normalized Transformer - tricks to keep the activations constrained, improves training convergence; from NVIDIA. Was poin….
0
82
0