Model save

README.md

@@ -34,12 +34,12 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 2e-05
-- train_batch_size: 13
+- train_batch_size: 11
 - eval_batch_size: 16
 - seed: 42
 - distributed_type: multi-GPU
 - gradient_accumulation_steps: 4
-- total_train_batch_size: 52
+- total_train_batch_size: 44
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - num_epochs: 1

adapter_config.json

@@ -16,10 +16,10 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
+    "k_proj",
     "o_proj",
-    "v_proj",
     "q_proj",
-    "k_proj"
+    "v_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

runs/Feb04_00-12-57_r2d2-devbox-fqvnh/events.out.tfevents.1738627994.r2d2-devbox-fqvnh.18919.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e6c71c1bd6d56ac3b484f13650395978b6f1414f9e2ca85b25798822f96f4a7
+size 4703

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:13bbc3463190dd66ef2467e45eed5bd6ee52a065a550b44c520b532aa53223de
+oid sha256:5b027ae1527c5de80c84f0d58eb362f13779af620421bc086e44c3e04992c32d
 size 5816