Training in progress, step 16

adapter_config.json

@@ -19,14 +19,14 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "w2",
-    "o_proj",
-    "w3",
     "w1",
+    "w3",
     "q_proj",
-    "gate",
+    "o_proj",
+    "k_proj",
     "v_proj",
-    "k_proj"
+    "w2",
+    "gate"
   ],
   "task_type": "CAUSAL_LM",
   "use_rslora": false

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e58fe9bba85719da247528ba0d3ece1062e7c4ab560a4f42f829653f9066b113
+oid sha256:517453a50eb7ad83390d805a80b868a81fb231d12e9ee38a6f6b88918c63920e
 size 969174880

config.json

@@ -21,9 +21,9 @@
   "quantization_config": {
     "_load_in_4bit": true,
     "_load_in_8bit": false,
-    "bnb_4bit_compute_dtype": "float32",
-    "bnb_4bit_quant_type": "fp4",
-    "bnb_4bit_use_double_quant": false,
+    "bnb_4bit_compute_dtype": "bfloat16",
+    "bnb_4bit_quant_type": "nf4",
+    "bnb_4bit_use_double_quant": true,
     "llm_int8_enable_fp32_cpu_offload": false,
     "llm_int8_has_fp16_weight": false,
     "llm_int8_skip_modules": null,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:06e79aedcf643534d926bfcff5634a06295a3865665c50db81d587f17849b728
+oid sha256:b75e67a82876c3f6341e84affb0de9103b7718bc9fc49cd3bdc190851350dfdb
 size 5688