Merge remote-tracking branch 'origin/staging' into fix_testing_dpo

opentensor · Aug 25, 2023 · 64e9e82 · 64e9e82
2 parents 1a65224 + 6286e9c
commit 64e9e82
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 1 deletion.
diff --git a/requirements.txt b/requirements.txt
@@ -1,5 +1,5 @@
-transformers<=4.31.0
 bittensor>=5.2.1,<6.0.0
+transformers<=4.28.0
 wandb==0.15.3
 datasets==2.14.0
 plotly==5.14.1

diff --git a/tests/test_event.py b/tests/test_event.py
@@ -42,6 +42,7 @@ def test_event_from_dict_all_forward_columns_match(self):
             RewardModelType.nsfw.value: [1.0],
             RewardModelType.reciprocate.value: [1.0],
             RewardModelType.diversity.value: [1.0],
+            RewardModelType.dpo.value: [1.0],
             RewardModelType.rlhf.value: [1.0],
             RewardModelType.prompt.value: [1.0],
             RewardModelType.relevance.value: [1.0],
@@ -100,6 +101,7 @@ def test_event_from_dict_forward_no_reward_logging(self):
         assert event.nsfw_filter is None
         assert event.reciprocate_reward_model is None
         assert event.diversity_reward_model is None
+        assert event.dpo_reward_model is None
         assert event.rlhf_reward_model is None
         assert event.prompt_reward_model is None
         assert event.relevance_filter is None
@@ -141,6 +143,7 @@ def test_event_from_dict_forward_reward_logging_mismatch(self):
         assert event.nsfw_filter is None
         assert event.reciprocate_reward_model is None
         assert event.diversity_reward_model is None
+        assert event.dpo_reward_model is None
         assert event.rlhf_reward_model is None
         assert event.prompt_reward_model is None
         assert event.relevance_filter is None