From 8ecdf35bd36b55c5a7b48b05fea6ae50085682b0 Mon Sep 17 00:00:00 2001
From: wassname <git@wassname.org>
Date: Sat, 25 Nov 2023 05:48:17 +0800
Subject: [PATCH] learns the best yet

---
 justfile            |  3 +++
 research_journal.md | 12 ++++++++++++
 src/trainer.py      |  2 +-
 3 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/justfile b/justfile
index 5f3c137..4c75710 100644
--- a/justfile
+++ b/justfile
@@ -16,3 +16,6 @@ watch_latest:
     cd *([-1]) && \
     cd *([-1]) && \
     scripts/play.sh -e -r -h
+
+default: 
+    just --list
diff --git a/research_journal.md b/research_journal.md
index 492e009..ed75ea4 100644
--- a/research_journal.md
+++ b/research_journal.md
@@ -388,3 +388,15 @@ but we have tokens vs z
 Questions:
 - wait why are we just passing in "action_token" to the transformer and not obs? that must have obs in it right... right??? confirm
 - in iris-delta how did they pass everything in? I guess obs_prev was tokenized too? I think the slices are annoying so maybe I should just pass things seperatly
+
+# 2023-11-24 10:56:40
+
+If I unfreeze the whole transformer, it seem to learn the most obvious dynamics (the next latent space is the same as the last). 
+
+To summarize
+- with Qlora it didn't learn that
+- with unfrozen head it didn't
+- when training transformer and obs embedding together it did not (frozen llm embeddings)
+
+
+no it didn't work with tokenizer sep hmm
diff --git a/src/trainer.py b/src/trainer.py
index e73699a..c3d6399 100644
--- a/src/trainer.py
+++ b/src/trainer.py
@@ -149,10 +149,10 @@ class Trainer:
 
         if epoch > cfg_tokenizer.start_after_epochs:
             metrics_tokenizer = self.train_component(self.agent.tokenizer, self.optimizer_tokenizer, sequence_length=1, sample_from_start=True, **cfg_tokenizer)
-        self.agent.tokenizer.eval()
 
         if epoch > cfg_world_model.start_after_epochs:
             metrics_world_model = self.train_component(self.agent.world_model, self.optimizer_world_model, sequence_length=self.cfg.common.sequence_length, sample_from_start=True, tokenizer=self.agent.tokenizer, **cfg_world_model)
+        self.agent.tokenizer.eval()
         self.agent.world_model.eval()
 
         if epoch > cfg_actor_critic.start_after_epochs: