Model save

README.md

@@ -41,6 +41,7 @@ The following hyperparameters were used during training:
 - total_train_batch_size: 256
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
+- lr_scheduler_warmup_steps: 2
 - num_epochs: 4
 
 ### Training results
@@ -49,7 +50,8 @@ The following hyperparameters were used during training:
 
 ### Framework versions
 
-- PEFT 0.13.2
+- PEFT 0.13.0
 - Transformers 4.45.2
-- Pytorch 2.4.1
-- Tokenizers 0.20.1
+- Pytorch 2.4.0
+- Datasets 2.21.0
+- Tokenizers 0.20.0

adapter_config.json

@@ -20,13 +20,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "q_proj",
-    "o_proj",
     "down_proj",
-    "v_proj",
-    "up_proj",
     "k_proj",
-    "gate_proj"
+    "up_proj",
+    "o_proj",
+    "gate_proj",
+    "q_proj",
+    "v_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6367bd7f0fb3ab036311a0aab0599e1ed32039ee2f89d86153d7b2b9be39edba
+oid sha256:1a62fcaa260a8997f02feab49617d5deee6d1dec793e9405094db6fea534bf21
 size 1635969696

trainer_state.json

@@ -10,68 +10,68 @@
   "log_history": [
     {
       "epoch": 0.4854368932038835,
-      "grad_norm": 0.9188441634178162,
-      "learning_rate": 8.786407766990292e-05,
-      "loss": 0.8368,
+      "grad_norm": 1.1003823280334473,
+      "learning_rate": 8.794946550048592e-05,
+      "loss": 0.8551,
       "step": 250
     },
     {
       "epoch": 0.970873786407767,
-      "grad_norm": 0.6881112456321716,
-      "learning_rate": 7.572815533980583e-05,
-      "loss": 0.4246,
+      "grad_norm": 0.7180963158607483,
+      "learning_rate": 7.580174927113704e-05,
+      "loss": 0.4137,
       "step": 500
     },
     {
       "epoch": 1.4563106796116505,
-      "grad_norm": 0.6344935297966003,
-      "learning_rate": 6.359223300970875e-05,
-      "loss": 0.3804,
+      "grad_norm": 0.6523966789245605,
+      "learning_rate": 6.365403304178815e-05,
+      "loss": 0.3735,
       "step": 750
     },
     {
       "epoch": 1.941747572815534,
-      "grad_norm": 0.6141209006309509,
-      "learning_rate": 5.145631067961165e-05,
-      "loss": 0.3646,
+      "grad_norm": 0.6745087504386902,
+      "learning_rate": 5.150631681243926e-05,
+      "loss": 0.3583,
       "step": 1000
     },
     {
       "epoch": 2.4271844660194173,
-      "grad_norm": 0.5896692276000977,
-      "learning_rate": 3.9320388349514564e-05,
-      "loss": 0.3538,
+      "grad_norm": 0.6706854104995728,
+      "learning_rate": 3.9358600583090386e-05,
+      "loss": 0.345,
       "step": 1250
     },
     {
       "epoch": 2.912621359223301,
-      "grad_norm": 0.6480565667152405,
-      "learning_rate": 2.7184466019417475e-05,
-      "loss": 0.3462,
+      "grad_norm": 0.7672198414802551,
+      "learning_rate": 2.72108843537415e-05,
+      "loss": 0.333,
       "step": 1500
     },
     {
       "epoch": 3.3980582524271843,
-      "grad_norm": 0.6381719708442688,
-      "learning_rate": 1.5048543689320387e-05,
-      "loss": 0.3338,
+      "grad_norm": 0.7942991256713867,
+      "learning_rate": 1.5063168124392615e-05,
+      "loss": 0.3142,
       "step": 1750
     },
     {
       "epoch": 3.883495145631068,
-      "grad_norm": 0.610920250415802,
-      "learning_rate": 2.912621359223301e-06,
-      "loss": 0.3259,
+      "grad_norm": 0.7696407437324524,
+      "learning_rate": 2.915451895043732e-06,
+      "loss": 0.302,
       "step": 2000
     },
     {
       "epoch": 4.0,
       "step": 2060,
       "total_flos": 1.4445804612483994e+18,
-      "train_loss": 0.4179296539825143,
-      "train_runtime": 17434.171,
-      "train_samples_per_second": 30.248,
-      "train_steps_per_second": 0.118
+      "train_loss": 0.40855656966422366,
+      "train_runtime": 23649.5822,
+      "train_samples_per_second": 22.298,
+      "train_steps_per_second": 0.087
     }
   ],
   "logging_steps": 250,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7484e26a2443ab001dada28afc5e4336045ee6f01f36bb190af4ea081f0e6eb8
+oid sha256:0b5fb93126c6b5a18f8421a4eeddaac11f98a23d05e6f786996af63a656ca025
 size 5240