Training in progress, step 300

adapter_config.json

@@ -14,13 +14,13 @@
   "r": 64,
   "revision": null,
   "target_modules": [
-    "gate_proj",
-    "k_proj",
     "q_proj",
-    "up_proj",
-    "down_proj",
     "v_proj",
-    "o_proj"
+    "up_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "down_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ee2ec909333d3905200788230f22925c9ffad5b1eaa153c525a871e6167df588
-size 6264
+oid sha256:4d0551ad96bb3a47f40b27101c8057f425b3b65b4a89dd1b9631ee395474864c
+size 6200

training_config.json

@@ -59,7 +59,7 @@
   "raw_lora_target_modules": "all",
   "output_dir": "./outputs/",
   "per_device_train_batch_size": 8,
-  "do_eval": false,
+  "do_eval": true,
   "per_device_eval_batch_size": null,
   "gradient_accumulation_steps": 2,
   "eval_accumulation_steps": null,