Merge branch 'train-olmo-large' of https://github.com/allenai/LLM int…

…o train-olmo-large
allenai · Mar 28, 2024 · f50c375 · f50c375
2 parents d41527d + c0a638a
commit f50c375
Show file tree

Hide file tree

Showing 2 changed files with 14 additions and 1 deletion.
diff --git a/configs/mcli/mitchish70.yaml b/configs/mcli/mitchish70.yaml
@@ -52,7 +52,7 @@ command: |-
   scripts/train.py configs/mitchish70-s3.yaml \
     --run_name=mitchish70-002 \
     --wandb.group=mitchish70-official \
-    --load_path=s3://ai2-llm/checkpoints/OLMo-large/mitchish70-002/step34700-unsharded \
+    '--load_path=${path.last_checkpoint:${remote_save_folder}}' \
     --global_train_batch_size=1536 \
     --device_train_microbatch_size=3 \
     --time_limit=604800 \
@@ -64,3 +64,4 @@ command: |-
 #    --load_path=s3://ai2-llm/checkpoints/OLMo-large/mitchish70-002/step32050-unsharded \
 #    --load_path=s3://ai2-llm/checkpoints/OLMo-large/mitchish70-002/step32300-unsharded \
 #    --load_path=s3://ai2-llm/checkpoints/OLMo-large/mitchish70-002/step32300 \
+#    --load_path=s3://ai2-llm/checkpoints/OLMo-large/mitchish70-002/step34700-unsharded \
diff --git a/configs/mitchish70-s3.yaml b/configs/mitchish70-s3.yaml
@@ -189,6 +189,18 @@ evaluators:
   - label: mmlu_other_var
     type: downstream
 
+  - label: mmlu_stem_mc_5shot
+    type: downstream
+
+  - label: mmlu_humanities_mc_5shot
+    type: downstream
+
+  - label: mmlu_social_sciences_mc_5shot
+    type: downstream
+
+  - label: mmlu_other_mc_5shot
+    type: downstream
+
 data:
   pad_direction: right
   num_workers: 8