Skip to content

Conversation

puyuan1996
Copy link
Collaborator

No description provided.

puyuan1996 and others added 30 commits June 4, 2025 00:51
…g and configurable reconstruction loss mode (#355)

* v0.2.0

* polish(pu): add final_norm_option_in_encoder

* polish(pu): polish jericho configs

* tmp

* fix(pu): fix world model init bug when use pretrained_model

* tmp

* feature(xjy): add text regularization function

* feature(xjy): add decode text regularization option and related logs (#348)

* fix(xjy): fixed some bug and add a function to output the decoder's text

* fix(pu): fix _shift_right in decode loss

* fix(xjy): add decode text function and  decode_loss_mode option of reconstruction loss for jericho (#363)

* Standardized the format and fixed existing bugs

* resolved game_buffer bug and polished formatting

* polish(xjy): standardize decode text related code for jericho (#366)

* polish(xjy): delete unnecessary comments and translate CN comments into EN

* fix(xjy): merged latest main branch (#368)

* v0.2.0

* style(pu): use actions/upload-artifact@v3

* fix(pu): fix Union import in game_segment

* style(pu): use actions/upload-artifact@v4

* test(nyz): only upload cov in macos

* fix(pu): fix reanalyze_ratio compatibility with rope embed (#342)

* fix(pu): fix release.yml

* fix(pu): fix release.yml (#343)

* fix(pu): fix release.yml

* fix(pu): fix release.yml

* fix(pu): fix release.yml

* fix(pu): fix release.yml

* fix(pu): fix release.yml

* fix(pu): use actions/download-artifact@v2

* fix(pu): use actions/download-artifact@v4

* release v0.2.0

* fix(lkj): fix typo in customize_envs.md

* fix(pu): adapt atari and dmc2gym env to support shared_memory (#345)

* fix(pu): fix atari and dmc2gym env to support shared_memory

* tmp

* fix(pu): fix frame_stack_num default cfg in atari env

---------

Co-authored-by: puyuan <[email protected]>

* delete unnecessary comments and translate CN comments into EN

* delete unnecessary comment

---------

Co-authored-by: 蒲源 <[email protected]>
Co-authored-by: PaParaZz1 <[email protected]>
Co-authored-by: 蒲源 <[email protected]>
Co-authored-by: 林楷傑 <[email protected]>
Co-authored-by: puyuan <[email protected]>

* latest remove unnucessary comments

* fix(pu): fix compatibility

* polish(pu): polish readme and requirements

---------

Co-authored-by: puyuan <[email protected]>
Co-authored-by: xiongjyu <[email protected]>
Co-authored-by: PaParaZz1 <[email protected]>
Co-authored-by: 林楷傑 <[email protected]>
…ero (#372)

fix timestep and non-text-based games for muzero
…orical representation ranges (#387)

* feature(fir): controlled reward/value categorical representation

* scaling_transform.py correction
…el (#391)

* Qwen is tested as a policy in the jericho environment

* fixed the bug that bad reflection cannot be collected

* supports options for selecting encoder/decoder

* fixed a few bugs and standardized the format

* standardize the format again

---------

Co-authored-by: puyuan <[email protected]>
…ty, fix _reset_collect/eval, add adaptive policy entropy control
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants