finetune llama2 on git

stanford-crfm · Apr 23, 2024 · e441126 · e441126
1 parent 28f299a
commit e441126
Showing 1 changed file with 6 additions and 2 deletions.
diff --git a/config/llama2_7b_continued.yaml b/config/llama2_7b_continued.yaml
@@ -1,6 +1,10 @@
 data:
-  id: EleutherAI/pile
-  tokenizer: meta-llama/Llama-2-7b-hf
+  cache_dir: "gs://levanter-data/tokenized/git-llama2/"
+  train_urls:
+    - gs://levanter-data/pile-domains/github/{00..29}.jsonl.zst
+  validation_urls:
+    - gs://levanter-data/pile-domains/github/val.jsonl.zst
+  tokenizer: "meta-llama/Llama-2-70b-hf"
 model:
   type: llama
 initialize_from_hf: true