Training in progress, step 23

adapter_config.json

@@ -3,6 +3,8 @@
   "auto_mapping": null,
   "base_model_name_or_path": "codellama/CodeLlama-34b-Python-hf",
   "bias": "none",
+  "eva_config": null,
+  "exclude_modules": null,
   "fan_in_fan_out": null,
   "inference_mode": true,
   "init_lora_weights": true,
@@ -11,6 +13,7 @@
   "layers_to_transform": null,
   "loftq_config": {},
   "lora_alpha": 16,
+  "lora_bias": false,
   "lora_dropout": 0.05,
   "megatron_config": null,
   "megatron_core": "megatron.core",
@@ -23,13 +26,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "o_proj",
     "q_proj",
+    "up_proj",
+    "o_proj",
     "down_proj",
     "v_proj",
     "gate_proj",
-    "up_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:de4620bc5bb8e206b5bf7381de1ef30fa8097f90b77da9ecfb1ab047ee1d4e75
-size 3173444576
+oid sha256:e01f4bdc2db7cc3217b59761abf161494d43b5eb323647215b618e1076e0108e
+size 3200671160

config.json

@@ -1,4 +1,5 @@
 {
+  "_attn_implementation_autoset": true,
   "_name_or_path": "codellama/CodeLlama-34b-Python-hf",
   "architectures": [
     "LlamaForCausalLM"
@@ -7,6 +8,7 @@
   "attention_dropout": 0.0,
   "bos_token_id": 1,
   "eos_token_id": 2,
+  "head_dim": 128,
   "hidden_act": "silu",
   "hidden_size": 8192,
   "initializer_range": 0.02,
@@ -23,7 +25,7 @@
   "rope_theta": 1000000,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.42.4",
+  "transformers_version": "4.46.3",
   "use_cache": false,
   "vocab_size": 32004
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:16eb7312c18e89fa9c2e11dd0fe944fa58203e7127b843ce222c56fe5306e271
-size 7864
+oid sha256:7d8e99f39be421f9370a798bae77fe79ab37264580c72fe615711b95ec4b8a4e
+size 6328