Commit graph

801 commits

Author SHA1 Message Date
interstellarninja
a4cdf80e4a BaseEnvConfig subclass for experimental vars 2025-12-26 17:04:12 +05:45
interstellarninja
2aa950a5a8 Add MT-GRPO turn-level advantage environment
- Implement turn-level credit assignment following MT-GRPO paper
  - Custom reward computation: turn-level + outcome-level rewards
  - Per-token advantage assignment compatible with existing GRPO trainer
  - Configurable lambda parameter for turn/outcome advantage weighting
2025-12-26 17:04:12 +05:45
pre-commit-ci[bot]
d04f8c0ae7 [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-12-25 09:54:56 +00:00
teknium
8435371d80 linty 2025-12-25 09:54:11 +00:00
pre-commit-ci[bot]
269fb71713 [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-12-25 09:49:33 +00:00
teknium
9e9f1cd88e Merge branch 'port_many_evals' of https://github.com/NousResearch/atropos into port_many_evals 2025-12-25 09:48:50 +00:00
teknium
c871f6a56a fix eval ctx len 2025-12-25 09:48:47 +00:00
pre-commit-ci[bot]
6bb6a5976d [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-12-24 23:37:21 +00:00
teknium
85296c519e hopefully final linter fixes lol 2025-12-24 23:36:36 +00:00
pre-commit-ci[bot]
d932d9c03b [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-12-24 21:27:53 +00:00
JoeLi12345
3348e31a29 readme 2025-12-24 21:22:07 +00:00
JoeLi12345
2fd888cdb0 lcb coding rl environment 2025-12-24 20:57:49 +00:00
pre-commit-ci[bot]
67869c3a79 [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-12-24 11:05:18 +00:00
teknium
148333a23b Merge branch 'port_many_evals' of https://github.com/NousResearch/atropos into port_many_evals 2025-12-24 11:04:35 +00:00
teknium
abdda3978a more linter nonsense 2025-12-24 11:04:33 +00:00
pre-commit-ci[bot]
fbf1a26559 [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-12-24 10:54:58 +00:00
teknium
f18d46549d fix linter errors 2025-12-24 10:53:45 +00:00
pre-commit-ci[bot]
afab28dfa9 [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-12-24 10:48:24 +00:00
teknium
ef9c0c3699 Port many benchmarks into atropos 2025-12-24 10:23:16 +00:00
kit
6eb2e49618 added missing logprob 2025-12-24 00:05:10 +00:00
Tonny
a25e299c83
Update README.md 2025-12-22 21:53:28 +03:00
Tonny
8da2b5ae29
Update README.md 2025-12-22 21:50:53 +03:00
Tonny
e0b870f28e
Update README.md 2025-12-22 21:50:39 +03:00
Tonny
1761f08211
Update README.md 2025-12-22 21:50:11 +03:00
Tonny
40f3c1f7e7
Update README.md 2025-12-22 21:49:55 +03:00
Dakota
8ec5066998 add eval runner 2025-12-19 19:56:59 -06:00
pre-commit-ci[bot]
d8c83dd5de [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-12-15 17:52:31 +00:00
GHOryy5
4c4aba108c
Prevent hangs in kernel evaluation by bounding worker waits 2025-12-15 20:50:03 +03:00
Juli
b8f0ba2271
Update README.md 2025-11-20 10:15:08 +01:00
Juli
fc594360ff
Update README.md 2025-11-20 10:14:49 +01:00
Juli
6cbc704d40
Update README.md 2025-11-20 10:14:30 +01:00
Juli
98dc606a87
Update README.md 2025-11-20 10:14:01 +01:00
Juli
b255f0b3ae
Update README.md 2025-11-20 10:13:36 +01:00
Teknium
c5c8ca57dc
Merge pull request #278 from NousResearch/conversion_to_managedserver
Convert Environments to ManagedServer for Tinker Integrations
2025-11-14 12:56:44 -08:00
teknium
9034d4c78e convert answer format env to use managedserver 2025-11-14 10:21:24 +00:00
teknium
ae101ea8e4 convert bootcamp to use managedserver 2025-11-14 10:17:48 +00:00
teknium
8e851a5ad4 convert kernelbench env to use managedserver 2025-11-14 10:15:01 +00:00
teknium
c4ecc42139 convert pydantic schema env to use managed server 2025-11-14 10:09:43 +00:00
pre-commit-ci[bot]
8cc83db6ee [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-11-14 10:02:28 +00:00
teknium
653a8b4543 convert reasoning gym env to use managedserver 2025-11-14 10:01:49 +00:00
pre-commit-ci[bot]
77e14199ce [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-11-14 09:53:34 +00:00
teknium
46f05673aa convert instruct following env to use managedserver 2025-11-14 09:52:02 +00:00
pre-commit-ci[bot]
a4d81e36d1 [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-11-14 09:51:28 +00:00
teknium
6d6a02eb38 convert instruction following env to use managed server 2025-11-14 09:49:04 +00:00
teknium
4738fabd57 convert fundamentals prediction env to use managed server 2025-11-14 09:48:56 +00:00
teknium
ff46cfff44 convert letter_counting_environment to use managed server 2025-11-14 09:44:20 +00:00
pre-commit-ci[bot]
aae4432a58 [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-11-14 06:55:56 +00:00
teknium
76fec8b919 convert rlaif_server to managedserver 2025-11-14 06:53:16 +00:00
teknium
d8c68a93e3 convert tool_calling_server to managedserver 2025-11-14 06:48:07 +00:00
pre-commit-ci[bot]
0a3c15c7ad [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
2025-11-14 06:14:21 +00:00