Best model so far - eval_loss: 0.5032 -------------------------------------------------------- (Trained with Unsloth)

adapter_config.json

@@ -1,7 +1,7 @@
 {
   "alpha_pattern": {},
   "auto_mapping": null,
-  "base_model_name_or_path": "unsloth/Qwen2-7B-Instruct",
+  "base_model_name_or_path": "unsloth/gemma-1.1-2b-it",
   "bias": "none",
   "fan_in_fan_out": false,
   "inference_mode": true,
@@ -11,7 +11,7 @@
   "layers_to_transform": null,
   "loftq_config": {},
   "lora_alpha": 128,
-  "lora_dropout": 0.05,
+  "lora_dropout": 0.1,
   "megatron_config": null,
   "megatron_core": "megatron.core",
   "modules_to_save": [
@@ -22,13 +22,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "gate_proj",
     "up_proj",
-    "o_proj",
-    "down_proj",
     "v_proj",
-    "q_proj"
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "gate_proj",
+    "k_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:13d7145a7782881bac14be188dcf333e2abefb739484b81b972e7fe4981408e3
-size 1735970568
+oid sha256:6d84613a15020c2a788bb9b7d8b3ad1e0bee4bafacf2c3c61db1b40282ea3db6
+size 1362396360