@earakely

Software Engineer @ Scale AI

Joined March 2023
  • Tweets376
  • Following176
  • Followers288
14Media
We're making rapid iterations on this in the coming weeks. The goal is that REA can be run on your laptop to produce an RL env of your computer and the work you do on it.
Xiangyi Li
@xdotliOct 21
Genius idea @earakely
1
8
601
Was listening to open.spotify.com/episode/2e7… and I like what is said at 26:23 from the databricks chief ai scientist: “many of the environments I look at are 10s of pbs, go put that in a docker container. And make it something that is easy to modify and reset. This is a new systems problem”. TLDR the why we built agentenvframework.com at @scale_AI

Jonathan Frankle - Chief AI Scientist at Databricks

The Frontier with Brendan Foody · Episode

open.spotify.com
1
9
7,755
Edgar Arakelyan retweeted
ZenGPT@FazGPTOct 20
Replying to @morluto @earakely @scale_AI
this might just be the holy grail of RL
1
106
Edgar Arakelyan retweeted
morluto
@morlutoOct 20
lots of ai & data labs are already using REA to help scale and build better environments for their models so we made it even easier with rea environments! a huge thank you to the @earakely and the @scale_AI team for all the help
reatools@reatoolsOct 20
Building good RL environments is hard. We're introducing a way to turn anything into RL-environments with REA using the open source AgentEnv framework from @ScaleAILabs Any site, app, or game can now easily be turned into an env 😉
10
12
1
186
7,977
Edgar Arakelyan retweeted
morluto
@morlutoOct 20
everything is an environment = RL on everything 🤫
Edgar Arakelyan
@earakelyOct 20
Everything can be an RL environment now
4
2
1
35
3,763
Everything can be an RL environment now
reatools@reatoolsOct 20
Building good RL environments is hard. We're introducing a way to turn anything into RL-environments with REA using the open source AgentEnv framework from @ScaleAILabs Any site, app, or game can now easily be turned into an env 😉
4
3
1
84
12,329
Learn more at agentenvframework.com

AgentEnv Framework | Scale AI's open-source framework for RL environments

The open-source framework Scale AI uses to build RL environments: composable environments behind an MCP gateway, any agent in any sandbox, tasks as DAGs.

agentenvframework.com
3
228
Teaser for the upcoming AirOps RL Environment. A high fidelity environment testing whether a model can run an airline during a snowstorm.
1
16
11,688
Preview available on huggingface.co/datasets/eara…. In collaboration with @adithya_s_k & team. Built using agentenvframework.com

earakely-scale/AirOpsEnv-AgentEnv · Datasets at Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co
165
Edgar Arakelyan retweeted
Chen
@chenzhangrealOct 16
Replying to @earakely @ScaleAILabs
that virtual clock fix is so underrated. nothing worse than rl runs breaking because time drifted.
1
49
Replying to @earakely @scale_AI @huggingface @ben_burtenshaw
Try it: pip install agentenv-framework agentenv-hf agent-env hf run earakely-scale/PortSimEnv-AgentEnv It shows what the environment needs and asks before it runs anything. Open source from @ScaleAILabs: agentenvframework.com

AgentEnv Framework | Scale AI's open-source framework for RL environments

The open-source framework Scale AI uses to build RL environments: composable environments behind an MCP gateway, any agent in any sandbox, tasks as DAGs.

agentenvframework.com
1
2
121
AgentEnv on the Hub, beside Harbor, Verifiers, OpenEnv and NeMo Gym: huggingface.co/datasets?libr… PortSimEnv: huggingface.co/datasets/eara… AirOpsEnv: huggingface.co/datasets/eara… OpenCiv3: huggingface.co/datasets/eara… PortSimEnv builds on @adithya_s_k's original in FineEnvs.

Datasets compatible with the AgentEnv library – Hugging Face

Explore datasets powering machine learning.

huggingface.co
92
we got lamport clocks for agents before gta 6
Edgar Arakelyan
@earakelyOct 16
Three agents and eight servers ask what time it is, all at once, and every one gets the same answer. At @ScaleAILabs, we built this virtual clock into our open-source AgentEnv Framework so RL environments stay reproducible. Here's why it matters and how it works 🧵
1
3
218
Three agents and eight servers ask what time it is, all at once, and every one gets the same answer. At @ScaleAILabs, we built this virtual clock into our open-source AgentEnv Framework so RL environments stay reproducible. Here's why it matters and how it works 🧵
5
5
2
32
11,731
more replies
4. One task step Putting a task on the virtual clock is one sync_env_clock step, right before the prompt. It arms the clock and syncs the servers that follow it, so virtual time starts with the agent.
1
1
75
Try it: pip install agentenv-framework A server follows the clock in a few lines. Full post: agentenvframework.com/blog/r… Code: github.com/scaleapi/agentenv… Written with @RazvanDuu, @thetututrain, Mateusz Cikowski and @liuying04. How do you handle time in your RL envs?
3
124
Edgar Arakelyan retweeted
Francis deSouza
@fdesouzaOct 15
The future of enterprise AI is multi-model, and enterprises deserve the freedom to choose. NEWS: We’re expanding that choice through our collaboration with @NVIDIA, post-training its open-weight Nemotron 3.5 Super VL for enterprise workflows. Here’s what it delivered for two of our customers: ✅ @Skydance: A complex finance workflow that reviews vendor credit memos, statements and invoices to find credits the company is owed. Post-trained Nemotron matched their current frontier model’s accuracy at 40x lower inference cost, with 5x faster response times. ✅ @EY_US: An M&A due diligence agent that finds and reconciles answers across hundreds of files in a virtual data room. Post-trained Nemotron matched EY’s current frontier model at about one-eighteenth of the inference cost, returning answers more than twice as fast. Put NVIDIA’s open-weight models to work for your business with Scale Post-Training. Thank you, @JensenHuang and the NVIDIA team, for the early access and collaboration. scale.com/partnership/nvidia
4
7
3
44
3,461
Load more