Training in progress, step 10

adapter_config.json

@@ -20,17 +20,17 @@
   "megatron_core": "megatron.core",
   "modules_to_save": null,
   "peft_type": "LORA",
-  "r": 32,
+  "r": 16,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "down_proj",
     "v_proj",
-    "up_proj",
     "o_proj",
+    "k_proj",
+    "down_proj",
+    "gate_proj",
     "q_proj",
-    "gate_proj"
+    "up_proj"
   ],
   "task_type": "CAUSAL_LM",
   "trainable_token_indices": null,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:427b18eebfba46e68b0176ba69144e8992325be503614254a6cb8655f9bd5738
-size 119802032
+oid sha256:f87f06c15de8ab0d5f6d946d12d16fa96c1ff11b22c8aed14c7a586447e9f4b1
+size 59934640

config.json

@@ -16,18 +16,18 @@
   "num_hidden_layers": 36,
   "num_key_value_heads": 2,
   "quantization_config": {
-    "_load_in_4bit": false,
-    "_load_in_8bit": true,
-    "bnb_4bit_compute_dtype": "float32",
-    "bnb_4bit_quant_storage": "uint8",
-    "bnb_4bit_quant_type": "fp4",
-    "bnb_4bit_use_double_quant": false,
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "bfloat16",
+    "bnb_4bit_quant_storage": "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,
     "llm_int8_threshold": 6.0,
-    "load_in_4bit": false,
-    "load_in_8bit": true,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
     "quant_method": "bitsandbytes"
   },
   "rms_norm_eps": 1e-06,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:01550cda20733eb72a8700a4093f519c746bb8aec0aff4aeb778b9779f3405c7
+oid sha256:092b720b8e020d691de5301eff908eac7e0231b2fcd711fbf27b0c0492ed64b4
 size 8888