Upload folder using huggingface_hub

adapter_config.json

@@ -1,38 +1,47 @@
 {
     "adapter_path": "jumbo_adapters",
-    "batch_size": 2,
+    "batch_size": 16,
     "config": null,
-    "data": "cleaned_data/",
+    "data": "data/",
     "fine_tune_type": "lora",
-    "grad_checkpoint": false,
-    "iters": 10000,
-    "learning_rate": 2e-05,
+    "grad_checkpoint": true,
+    "iters": 3000,
+    "learning_rate": 5e-05,
     "lora_parameters": {
         "keys": [
             "mlp.gate_proj",
             "mlp.down_proj",
             "self_attn.q_proj",
             "mlp.up_proj",
-            "self_attn.o_proj",
             "self_attn.v_proj",
             "self_attn.k_proj"
         ],
-        "rank": 8,
-        "alpha": 8,
-        "dropout": 0.05,
+        "rank": 64,
+        "alpha": 64,
+        "dropout": 0.1,
         "scale": 16.0
     },
-    "lr_schedule": null,
-    "max_seq_length": 2048,
-    "model": "ministral/Ministral-3b-instruct",
-    "num_layers": 14,
+    "lr_schedule": {
+        "name": "cosine_decay",
+        "warmup": 500,
+        "warmup_init": 1e-07,
+        "arguments": [
+            1e-05,
+            500,
+            1e-07
+        ]
+    },
+    "max_seq_length": 512,
+    "model": "Qwen/Qwen2.5-3B",
+    "num_layers": 36,
     "resume_adapter_file": null,
-    "save_every": 100,
-    "seed": 0,
-    "steps_per_eval": 200,
+    "save_every": 50,
+    "seed": 24,
+    "steps_per_eval": 50,
     "steps_per_report": 10,
     "test": true,
-    "test_batches": 500,
+    "test_batches": 200,
+    "testfile": "/Users/dutingzhen/PycharmProjects/finetuning/data/test.jsonl",
     "train": true,
-    "val_batches": 25
+    "val_batches": 50
 }

adapters.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:004a9e19d8d2ae1a2e8a0309eb34c33938e4488a964eaee886f71d9b049f8ee7
-size 36721232
+oid sha256:b692f731342b23ea987aa4b3a09eb989b220062b56e59bdf034bdbbd424eef90
+size 441236095