Commit graph

60 commits

Author SHA1 Message Date
sam-paech
e351aa3841 savegame fix + chart updates 2025-07-05 00:27:02 +10:00
sam-paech
7edc7c465f fix prompt errors & add per-power prompt dir functionality 2025-07-04 11:31:57 +10:00
sam-paech
a241e34496 fix to respect model ids passed as args when resuming 2025-07-03 09:59:47 +10:00
sam-paech
1cb24f1884 set minimalist prompts as harness defaults 2025-07-03 07:27:48 +10:00
AlxAI
7b633a0ec8 making phase summaries and unformatted prompts params in lm_game as well as making the models changeable in utils 2025-06-30 12:03:37 +02:00
AlxAI
861a5a222f fixed un-needed changes 2025-06-30 09:32:26 +02:00
sam-paech
b5a84867a1 add order history 2025-06-29 01:53:03 +10:00
sam-paech
ebf26cf8a6 add simplified prompts 2025-06-27 14:42:05 +10:00
sam-paech
840c6b0ad9 add experiment runner 2025-06-22 18:05:07 +10:00
sam-paech
7d50b31e34 add resuming support + critical state analysis mode 2025-06-22 14:41:29 +10:00
sam-paech
a4855caaae add new args to main script: max_tokens & max_tokens_per_model 2025-06-20 05:58:47 +10:00
sam-paech
b405cf30c7 revamp diary consolidation; simplify negotiations prompt 2025-06-19 15:39:20 +10:00
AlxAI
9b24abef53 webhooks 2025-06-02 20:33:36 -04:00
AlxAI
4b92dd5af0 updating analysis with lie detection (it's not great yet) 2025-05-24 20:44:23 -04:00
AlxAI
742e260464 fixing eliminated powers 2025-05-21 21:27:49 -07:00
AlxAI
9322ada62b analyze moments, run big models well 2025-05-20 20:04:19 -07:00
AlxAI
f36d5672ea consolidation of diary! 2025-05-18 19:51:34 -04:00
AlxAI
db827de273 first diary 2025-05-18 17:23:47 -04:00
AlxAI
c50ac85758 analyze game moments attempt 1 2025-05-17 22:28:22 -04:00
AlxAI
bfcb9ce401 XML didn't work 2025-05-14 22:03:13 -04:00
AlxAI
b4bc48337e fix relationships, good to go backend 2025-05-12 13:27:57 -04:00
AlxAI
3a935c0491 fixed diary 2025-05-12 10:37:34 -04:00
AlxAI
94313c16d9 fix relationships 2025-05-11 22:19:20 -04:00
AlxAI
0bd6428729 BIG UPDATES logging everything, better structure of moves, everything runs fast af 2025-05-11 19:10:18 -04:00
sam-paech
0c7b0157b5 add private diary summaries 2025-05-11 18:38:37 +10:00
AlxAI
53e6a8fd6a saving logs 2025-05-05 10:46:34 -04:00
AlxAI
1dc25702b6 relationships work! Everything is ready for big runs 2025-05-04 11:21:51 -04:00
AlxAI
e6ba8bfbf1 summaries working not statistical summary though 2025-04-30 23:16:57 -04:00
AlxAI
02118dc98b async!! 2025-04-29 22:28:53 -04:00
AlxAI
eeddac0ef9 its working! 2025-04-29 20:51:46 -04:00
AlxAI
6e5079fa02 working with agent, relationships, and goals (seemingly) 2025-04-09 22:24:10 -07:00
AlxAI
70f4438b2e state! 2025-04-07 17:25:12 -07:00
AlxAI
0242d7446b Revert "Merge branch 'main' into animation"
This reverts commit d7f93f587a, reversing
changes made to d505c7ea6c.
2025-03-04 20:31:14 -08:00
AlxAI
d7f93f587a
Merge branch 'main' into animation 2025-03-04 20:26:35 -08:00
Oam Patel
1f8ac5ae20 add optional planning phase 2025-02-27 02:10:48 +00:00
AlxAI
cb04ad0be5 attempt at fixing recursive summarization 2025-02-25 06:55:21 -08:00
AlxAI
eb3de01956 dramatically improving logging thanks to new 3.7sonnet cursor agent mode 2025-02-24 15:49:37 -08:00
AlxAI
b54a8252d6 fix convoy first attempt at summaries 2025-02-23 18:18:47 -08:00
AlxAI
2693b01014 Lots of improvements to prompting putting the right information in for negotiation and phase summaries - CONVOYS BROKEN RN 2025-02-23 11:18:37 -08:00
AlxAI
6b0863cb5b dramatically simplify phase summary 2025-02-20 22:24:24 -08:00
AlxAI
8f61ba06b3 fixed system prompt for summary , made improvements and debugging for summaries too. Much can be optimized still 2025-02-20 18:22:53 -08:00
AlxAI
72327cfb22 Randomization for powers and models + enhance order instructions
Also improved plotting to show model + power
2025-02-20 15:58:41 -08:00
AlxAI
b886fd7bfc with phase summaries and country prompts 2025-02-19 21:26:54 -08:00
Oam Patel
ed10ae4b15 multithreaded negotiation + orders in context + fix GameHistory class 2025-02-18 23:10:23 +00:00
Oam Patel
e3db81f00e no press + context includes orders and results 2025-02-18 19:29:04 +00:00
Oam Patel
2f5ec5cdf7 ruff 2025-02-17 05:20:33 +00:00
Oam Patel
a0404acff9 multi message sending + further planning 2025-02-17 05:20:06 +00:00
Oam Patel
67e06f1e40 big refactor + big prompt changes + context prompt 2025-02-17 04:02:25 +00:00
Oam Patel
1b042cf6c6 script + error logging for order decoding 2025-02-16 01:27:16 +00:00
Oam Patel
a1aed6674f refactor 2025-02-16 00:24:57 +00:00