Mehmet Hamza Erol

17 posts

Mehmet Hamza Erol

Mehmet Hamza Erol

@mhamzaerol

Katılım Ekim 2021
124 Takip Edilen59 Takipçiler
Mehmet Hamza Erol retweetledi
Xiaokang(Koe) Ye
Xiaokang(Koe) Ye@koe_ye40329·
Super excited to introduce SimWorld Studio!🏭 With built-in coding agent SimCoder🦞, you can vibe-code physical 3D scenes in Unreal Engine -- then train embodied agents directly inside the generated world.🦾 From environment generation to agent training, end-to-end. 🚀
SimWorld@simworld_ai

Environment generation is the missing scaling axis for embodied AI. Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D env where agents act, fail & learn. Env-agent co-evolvution improves navigation success 50% → 90%. From a prompt, our SimCoder writes code to automatically build an interactive world. Agents train inside it. And their performance shapes the next world.

English
31
67
999
2.1M
Mehmet Hamza Erol retweetledi
SimWorld
SimWorld@simworld_ai·
Environment generation is the missing scaling axis for embodied AI. Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D env where agents act, fail & learn. Env-agent co-evolvution improves navigation success 50% → 90%. From a prompt, our SimCoder writes code to automatically build an interactive world. Agents train inside it. And their performance shapes the next world.
English
7
34
213
2.3M
Mehmet Hamza Erol retweetledi
Batu El
Batu El@elb4tu·
1/ Today I’m presenting our paper cost-of-pass at #ICLR2026! How does the cost of solving cognitive tasks change with innovations in LLMs? We introduce cost-of-pass and show something that looks like Moore's law for the cost of cognitive labor.
English
3
4
13
1.7K
Mehmet Hamza Erol retweetledi
Together AI
Together AI@togethercompute·
New from Together Research: LLMs can fix query plans your database optimizer gets wrong. Up to 4.78x faster. Cost estimators fail when they miss semantic correlations: wrong join order, wrong access path, cascading errors. DBPlanBench feeds DataFusion's physical operator graph to an LLM, which patches the plan directly instead of regenerating it from scratch. On TPC-H / TPC-DS: → 4.78x peak speedup → 60.8% of queries improved >5% → Build memory: 3.3 GB → 411 MB Optimize on small-scale data, transfer to production.
Together AI tweet media
English
2
7
38
6.1K
Mehmet Hamza Erol retweetledi
James Zou
James Zou@james_y_zou·
Cost-of-Pass will be presented in #iclr2026! We develop an economic framework to quantify the contributions of different LLMs💰. Great job by @mhamzaerol @elb4tu Mirac Suzgun and @mertyuksekgonul!
Mehmet Hamza Erol@mhamzaerol

How much does a correct answer from an LM cost? How much has AI lowered the cost of solving problems? Meet Cost‑of‑Pass: An Economic Framework for Evaluating LMs! Cost‑of‑Pass = expected $ for one correct answer. Frontier Cost‑of‑Pass = cheapest route: an LM or a human expert.

English
0
3
11
3.9K
Mehmet Hamza Erol retweetledi
James Zou
James Zou@james_y_zou·
We found a troubling emergent behavior in LLM. 💬When LLMs compete for social media likes, they start making things up 🗳️When they compete for votes, they turn inflammatory/populist When optimized for audiences, LLMs inadvertently become misaligned—we call this Moloch’s Bargain
James Zou tweet media
English
845
2K
9.7K
1.3M
Mehmet Hamza Erol
Mehmet Hamza Erol@mhamzaerol·
Overall, Cost-of-Pass offers a grounded economic lens on AI progress: it benchmarks models, spotlights which classes or techniques drive cost-effective improvements, and offers a practical way to track real-world usability and economic value of AI innovations.
Mehmet Hamza Erol tweet media
English
1
1
2
548
Mehmet Hamza Erol
Mehmet Hamza Erol@mhamzaerol·
How much does a correct answer from an LM cost? How much has AI lowered the cost of solving problems? Meet Cost‑of‑Pass: An Economic Framework for Evaluating LMs! Cost‑of‑Pass = expected $ for one correct answer. Frontier Cost‑of‑Pass = cheapest route: an LM or a human expert.
Mehmet Hamza Erol tweet media
English
4
24
71
23.8K

Keşfet

@elb4tu @mertyuksekgonul @ApacheDataFusio @federicobianchy @GreCo_CiRo @james_y_zou @suzgunmirac @elonmusk