Commit graph

508 commits

Author SHA1 Message Date
David van Vliet
208c15a4c1 further UI improvements with character img 2025-05-18 16:41:23 -07:00
Jake Boggs
cf9429f47f Move to punchlines folder 2025-05-18 19:39:33 -04:00
arihanv
94825011a0 Convert FOB submodule to regular folder 2025-05-18 16:36:28 -07:00
Roshan Sanjeev
0b2baffab5 ExamCraft: Working adaptive teacher environment with demo 2025-05-18 16:33:01 -07:00
Earl Potters
3420a0f988 Merge branch 'main' of https://github.com/Slyracoon23/atropos 2025-05-18 23:31:12 +00:00
Dylan Anderson
1489b02bbb add a patient 2025-05-18 16:25:03 -07:00
Shannon Sands
89386598d8 trying different models,no completions? 2025-05-18 16:22:00 -07:00
97hongjun
128dce55bc fix bugs in cat_server 2025-05-18 16:20:42 -07:00
Shannon Sands
0e84f2f47a fix up score 2025-05-18 16:17:14 -07:00
arihanv
48941b455c update: sandbox 2025-05-18 16:11:00 -07:00
David van Vliet
816489c9bd new UI and openrouter integration 2025-05-18 16:09:58 -07:00
Shannon Sands
98ce4224ff commented out server 2025-05-18 16:08:42 -07:00
tsadpbb
189d7a3dd1
Add dataset (#3) 2025-05-18 16:01:54 -07:00
Joshua Jerin
b298a0eeb6 train 2025-05-18 16:00:43 -07:00
arihanv
f70103c781 update: sandbox 2025-05-18 16:00:35 -07:00
Adam Blumenfeld
1f5ace6113 progress on README.md with packages and shared library considerations 2025-05-18 23:00:03 +00:00
Adam Blumenfeld
32274ae20f copy over grpo example and modify run section 2025-05-18 22:59:44 +00:00
olivier
e91de77112 first experiment on lambda 2025-05-18 22:58:34 +00:00
97hongjun
6c18d04bce playing around with cat_server.py 2025-05-18 15:52:52 -07:00
Earl Potters
5257b56063 Refine DynastAI choice history display and reward scoring
- Updated the prompt to clarify that previous choices are displayed in order.
- Removed numbering from character prompts for improved readability.
- Adjusted reward scoring to penalize with -100.0 instead of -1.0 for no reward.
2025-05-18 22:50:57 +00:00
Jake Boggs
9b920bc7c6
New zip file 2025-05-18 18:49:58 -04:00
Jake Boggs
45ee84476a Update device 2025-05-18 18:48:54 -04:00
Alexander Speicher
02ff663ebe Patient Doctor Loop 2025-05-18 15:45:59 -07:00
Alexander Speicher
4e7583be43 base changes 2025-05-18 15:45:59 -07:00
Adam Blumenfeld
c10ebe4b85 add pyrender_utils with test 2025-05-18 22:41:27 +00:00
Earl Potters
ad21f178a1 Update choice history display in DynastAI prompt formatting
- Modified the prompt generation to include all choices made by the player instead of limiting to the last three.
- Enhanced the formatting to ensure clarity by stripping existing numbering from character prompts.
- Improved readability of the decision and effects displayed for each choice in the history.
2025-05-18 22:35:12 +00:00
Joshua Jerin
7e1de80695 add jsonl file 2025-05-18 15:27:37 -07:00
Earl Potters
790e3848f5 Enhance DynastAI evaluation and prompt formatting with detailed debug logging
- Added debug print statements throughout the evaluation process to track the processing of test cards, input data, and generated prompts.
- Improved the `format_prompt` method to include detailed logging of kingdom state and choice history.
- Enhanced logging for training data selection to aid in debugging and understanding scenario generation.
2025-05-18 22:23:51 +00:00
Jake Boggs
2f3547bd82 Make more concise 2025-05-18 18:12:46 -04:00
Dylan Anderson
5351bcccea Get a template working 2025-05-18 15:06:46 -07:00
Jake Boggs
2b5910a19f Add impact section 2025-05-18 18:02:11 -04:00
Earl Potters
6f1a4f9a0b Merge branch 'main' into dynastai-reward-env 2025-05-18 21:59:52 +00:00
Earl Potters
7101aad190 Add DynastAI integration documentation and setup scripts
- Introduced `ATROPOS_INTEGRATION.md` for detailed instructions on using DynastAI with Atropos.
- Added `INSTALL_AND_RUN.md` to guide users through installation and running the game.
- Created `run_dynastai.py` for a simplified testing experience without full Atropos setup.
- Implemented `setup.py` to manage dependencies and ensure compatibility.
- Updated `requirements.txt` to include additional dependencies and version constraints.
- Enhanced `README.md` with new sections on installation, running the game, and integration with Atropos.
- Added installation verification script `verify_install.py` to check for required packages.
- Updated game logic to support local card generation and improved API integration.
- Enhanced web interface with new features for user interaction and game metrics display.
2025-05-18 21:59:12 +00:00
David van Vliet
38049522c8 formatting changes so that the game works 2025-05-18 14:52:32 -07:00
Adam Blumenfeld
96fd81daab add __init__.py 2025-05-18 21:50:12 +00:00
Jake Boggs
7efcb965bd
Add zip file 2025-05-18 17:45:44 -04:00
Josh
8ff2b02ce0 Working POC 2025-05-18 14:44:16 -07:00
Jake Boggs
09eed6584d Add example output 2025-05-18 17:43:38 -04:00
Jonah Philion
98e103d2f2 copy over gsmk 2025-05-18 14:42:58 -07:00
Earl Potters
b9a1940d12 Enhance evaluation process in DynastAI with detailed logging
- Added print statements to track the evaluation process, including the number of test scenarios and accuracy after evaluation.
- Improved JSON validation with additional logging for missing fields and invalid data types.
- Introduced counters for valid and invalid scenario generations during scoring.
- Enhanced feedback on scenario generation and processing errors for better debugging.
2025-05-18 21:23:33 +00:00
Jake Boggs
71a87edd50 Initial version 2025-05-18 17:21:16 -04:00
Joshua Jerin
e018440d66 initial - env 2025-05-18 14:16:01 -07:00
arihanv
fa42662c54 feat: optimizer evaluator 2025-05-18 14:15:32 -07:00
Earl Potters
fba2a42bb7 Enhance DynastAI scenario generation with structured point system and history tracking
- Updated point system guidelines to restrict values for Piety, Stability, Power, and Wealth to a range of -20 to 20.
- Introduced a category field for scenario classification.
- Added examples for scenario generation to improve clarity.
- Implemented choice history tracking to maintain a record of player decisions and their effects on the kingdom's state.
- Enhanced prompt formatting to include current kingdom state and recent choices for better context in scenario generation.
2025-05-18 21:13:13 +00:00
Alexander Speicher
1e7c3d7cc1 init 2025-05-18 13:55:40 -07:00
Adam Blumenfeld
3fde5cbda8 update gitignore 2025-05-18 20:52:54 +00:00
Olivier D'Ancona
366f82d5bc Merge branch 'main' of github.com:ODAncona/NousWhiteHouse 2025-05-18 13:41:24 -07:00
Olivier D'Ancona
0fbb112eec initialized with grpo 2025-05-18 13:41:01 -07:00
Josh
94038876f4 Add tokenizer. Fix typing 2025-05-18 13:32:59 -07:00
Josh
659247fc00 Fix environment issues. Safely ran python3 accessibility_env.py --help 2025-05-18 13:07:08 -07:00