Training in progress, step 50

adapter_config.json

@@ -20,13 +20,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "gate_proj",
+    "down_proj",
     "k_proj",
     "v_proj",
+    "gate_proj",
     "q_proj",
-    "o_proj",
     "up_proj",
-    "down_proj"
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1849ebd18594277dc7c334f19039234e27241446b681d5a0e21289c9b21b157c
+oid sha256:18984a389dec3802f279ab2d5ca6adda74180b8b37faa46f34ddca3ec3510b33
 size 83945296

llama3_lora_sft.yaml

@@ -32,7 +32,7 @@ load_best_model_at_end: true
 per_device_train_batch_size: 1
 gradient_accumulation_steps: 16
 learning_rate: 1.0e-4
-num_train_epochs: 10.0
+num_train_epochs: 5.0
 lr_scheduler_type: cosine
 warmup_ratio: 0.1
 bf16: true

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b738b9bee55cd0da1b56f1cc59cf441a167494fc4ba6138504da1fb7791fe507
+oid sha256:460ddaa0054fb58db1c7fc01e35d2b66fe1d6f8565617e2468ee89b2d4efa51b
 size 5624