56 commits

Author	SHA1	Message	Date
Isadora White	93db8b664c	fixing construction tasks	2025-03-22 17:13:14 -05:00
Isadora White	57af4f13cc	adding back logging	2025-03-22 15:17:08 -05:00
MaxRobinsonTheGreat	6dc5c6401a	Merge branch 'main' into cleanup	2025-03-20 16:45:20 -05:00
Isadora White	e3d61ceead	changing task path	2025-03-20 15:36:45 -05:00
Isadora White	7bf97660eb	added exp name to prompter.js and splitting around think tokens	2025-03-19 23:52:08 -05:00
Isadora White	6f8027b86f	add additional logging of config settings	2025-03-19 23:34:22 -05:00
Isadora White	bb50486e81	add new options and better logging	2025-03-19 23:29:11 -05:00
Isadora White	94faf8f82a	Merge branch 'main' of https://github.com/icwhite/mindcraft	2025-03-17 13:11:21 -07:00
Isadora White	15ef88aa61	silly deletion issue	2025-03-17 13:11:14 -07:00
Isadora White	dd5dd72381	Merge pull request #7 from icwhite/add-ollama-evaluation ... Add ollama evaluation	2025-03-17 12:38:32 -07:00
hlillemark	5a230e707e	Edit eval script to work with ollama model	2025-03-17 11:53:31 -07:00
Isadora White	fe4e75612d	one more try catch block around starting the server	2025-03-16 22:10:32 -07:00
Isadora White	2e4974ddd8	dynamic logging	2025-03-16 21:56:08 -07:00
Isadora White	192d92bc06	fix restart and the issue where the server sometimes doesn't get deleted properly	2025-03-16 20:31:30 -07:00
Isadora White	19bedf0593	making eval script more robust to server randomly crashing	2025-03-16 18:50:34 -07:00
Isadora White	1ccba3a4b5	new train, test, dev tasks and new analysis files	2025-03-16 17:55:05 -07:00
Isadora White	125aa73d6c	adding blocked actions	2025-03-14 18:51:41 -07:00
Isadora White	d399f8e214	blocked actions	2025-03-14 18:49:33 -07:00
Isadora White	8c78398056	fix single agent issue	2025-03-14 15:19:26 -07:00
Isadora White	dc3322b518	more /op commands	2025-03-14 00:14:37 -07:00
Isadora White	d8e933a25d	small fix to loading with cheats	2025-03-13 21:31:16 -07:00
Isadora White	406ebe6072	add url option to evaluation script	2025-03-09 23:14:26 -07:00
Isadora White	5103cd82eb	evaluation script small update	2025-03-09 22:51:00 -07:00
Isadora White	5b95b2f816	confirming correct profile path	2025-03-09 13:16:57 -07:00
Isadora White	2f80b65d42	fixing small bugs related to single agent support	2025-03-09 13:01:54 -07:00
Isadora White	3576c6fb07	Merge branch 'main' into main_save_03-07-2025	2025-03-09 12:02:58 -07:00
Isadora White	e9e8cf3c88	single agent task support in evaluation script	2025-03-09 12:02:48 -07:00
Isadora White	b75d941d97	smol changes	2025-03-08 19:24:13 -08:00
Isadora White	7492582ed9	changed eval script slightly	2025-03-07 17:29:43 -08:00
Mehul Maheshwari	dda1ba200f	better readability for incoming devs. revised blueprint generation for ladder fix. cheats enabled for jill_0 and andy_0.	2025-03-06 16:16:02 -08:00
Isadora White	9aefb4676a	updating port number for vllm	2025-03-05 15:53:43 -08:00
Isadora White	7e7f893cf3	fix evaluation script bug	2025-03-05 10:10:24 -08:00
Isadora White	6e0d7e1eaf	change evaluation_script to allow for new world names	2025-03-04 22:08:04 -08:00
Isadora White	34145168dc	refactoring changes	2025-03-04 12:09:23 -08:00
Isadora White	e3ec9d34b4	fixing merge related small bugs	2025-03-04 11:54:09 -08:00
mmaheshwari2	2036288e13	tested eval script, changed model to 4o-mini	2025-03-04 10:28:11 -08:00
Isadora White	44fc1b4618	evaluation script for vllm	2025-03-03 06:10:52 +00:00
Isadora White	37b1fc0bed	increase timeout length for adding bot to world for the first time	2025-03-01 21:37:20 -08:00
Isadora White	af79c78fbb	fixing evaluation script to actually add bots as op and add new models	2025-03-01 19:21:35 -08:00
Isadora White	ae39028d3b	longer sleeps, early breaking for scenarios where there is only one agent	2025-02-28 18:31:19 -08:00
Isadora White	39cec7cf82	changed the checking if complete cycle to be more frequent and updated the collab_profile	2025-02-28 16:59:56 -08:00
Isadora White	7cafc254d1	making the default to load in the collaborative profiles	2025-02-23 21:11:08 -08:00
Isadora White	2da97b5607	adding a mechanism to add environment variables to the keys.json automatically	2025-02-23 18:55:13 -08:00
Isadora White	8a75d8a78e	changing the give to player command to account for an edge case where the players are too close together and moving away takes time	2025-02-22 17:53:42 -08:00
Isadora White	719b72da9e	set up to use s3 logging instead of wandb	2025-02-21 17:02:21 -08:00
Isadora White	d4565aa68c	small fixes, the items were being given twice to the agents on initialization and accounting for blocked_actions not being in the task file	2025-02-20 21:45:29 -08:00
Isadora White	7a19f34e22	Merge branch 'main' into evaluation_parallelization	2025-02-18 18:34:46 -08:00
Isadora White	aad19d616c	fixed evaluation script to allow for parallel worlds again	2025-02-18 16:39:31 -08:00
Isadora White	fb5d95debe	fixed the issue with garbling commands by instead putting the commands in a bash script and running them that way	2025-02-17 17:25:12 -08:00
Isadora White	bc15700196	fixed wandb logging	2025-02-15 12:02:44 -08:00