Model save

README.md

@@ -33,17 +33,18 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 1e-06
-- train_batch_size: 8
-- eval_batch_size: 4
+- train_batch_size: 4
+- eval_batch_size: 2
 - seed: 42
 - distributed_type: multi-GPU
 - num_devices: 2
+- gradient_accumulation_steps: 2
 - total_train_batch_size: 16
-- total_eval_batch_size: 8
+- total_eval_batch_size: 4
 - optimizer: Adam with betas=(0.9,0.95) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.01
-- num_epochs: 1.0
+- num_epochs: 2.0
 
 ### Training results
 
@@ -52,7 +53,7 @@ The following hyperparameters were used during training:
 ### Framework versions
 
 - PEFT 0.12.0
-- Transformers 4.44.0
+- Transformers 4.40.0
 - Pytorch 2.1.2
 - Datasets 2.20.0
 - Tokenizers 0.19.1

adapter_config.json

@@ -2,7 +2,7 @@
   "alpha_pattern": {},
   "auto_mapping": {
     "base_model_class": "MiniCPMV",
-    "parent_library": "transformers_modules.openbmb.MiniCPM-V-2_6.ad23470e5e3b45a91ab9b253530c8b9c8cf36759.modeling_minicpmv"
+    "parent_library": "transformers_modules.openbmb.MiniCPM-V-2_6.b6f4084e52b67fceb71d3b963cb481f515a7e101.modeling_minicpmv"
   },
   "base_model_name_or_path": "openbmb/MiniCPM-V-2_6",
   "bias": "none",
@@ -19,7 +19,8 @@
   "megatron_core": "megatron.core",
   "modules_to_save": [
     "embed_tokens",
-    "resampler"
+    "resampler",
+    "vpm"
   ],
   "peft_type": "LORA",
   "r": 64,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:55d57698119acf4285cd0482969db532ee6b89a2a4f12dbb14e12983bcb2890b
-size 1305154552
+oid sha256:ef5bd84ec9d072204570d549b2d959789a1a5acb9dbf7fb3fb70843598f006d5
+size 2140797224

trainer_state.json

@@ -1,140 +1,226 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
+  "epoch": 2.0,
   "eval_steps": 1000,
-  "global_step": 72,
+  "global_step": 144,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.06944444444444445,
-      "grad_norm": 4.683212757110596,
+      "grad_norm": 8.71909236907959,
       "learning_rate": 1e-06,
-      "loss": 2.5448,
+      "loss": 2.6593,
       "step": 5
     },
     {
       "epoch": 0.1388888888888889,
-      "grad_norm": 4.721557140350342,
+      "grad_norm": 8.853459358215332,
       "learning_rate": 1e-06,
-      "loss": 2.4235,
+      "loss": 2.4766,
       "step": 10
     },
     {
       "epoch": 0.20833333333333334,
-      "grad_norm": 3.7502753734588623,
+      "grad_norm": 6.577568054199219,
       "learning_rate": 1e-06,
-      "loss": 2.5413,
+      "loss": 2.495,
       "step": 15
     },
     {
       "epoch": 0.2777777777777778,
-      "grad_norm": 4.788341045379639,
+      "grad_norm": 8.3203706741333,
       "learning_rate": 1e-06,
-      "loss": 2.7545,
+      "loss": 2.6105,
       "step": 20
     },
     {
       "epoch": 0.3472222222222222,
-      "grad_norm": 4.027582168579102,
+      "grad_norm": 7.732483863830566,
       "learning_rate": 1e-06,
-      "loss": 2.398,
+      "loss": 2.4744,
       "step": 25
     },
     {
       "epoch": 0.4166666666666667,
-      "grad_norm": 4.083437442779541,
+      "grad_norm": 6.6942548751831055,
       "learning_rate": 1e-06,
-      "loss": 2.737,
+      "loss": 2.5163,
       "step": 30
     },
     {
       "epoch": 0.4861111111111111,
-      "grad_norm": 3.840348720550537,
+      "grad_norm": 5.781284809112549,
       "learning_rate": 1e-06,
-      "loss": 2.6366,
+      "loss": 2.3854,
       "step": 35
     },
     {
       "epoch": 0.5555555555555556,
-      "grad_norm": 3.376926898956299,
+      "grad_norm": 7.311328887939453,
       "learning_rate": 1e-06,
-      "loss": 2.5638,
+      "loss": 2.4442,
       "step": 40
     },
     {
       "epoch": 0.625,
-      "grad_norm": 3.656587600708008,
+      "grad_norm": 6.254249572753906,
       "learning_rate": 1e-06,
-      "loss": 2.7568,
+      "loss": 2.2468,
       "step": 45
     },
     {
       "epoch": 0.6944444444444444,
-      "grad_norm": 2.5398614406585693,
+      "grad_norm": 7.6778669357299805,
       "learning_rate": 1e-06,
-      "loss": 2.3856,
+      "loss": 2.2565,
       "step": 50
     },
     {
       "epoch": 0.7638888888888888,
-      "grad_norm": 2.253296136856079,
+      "grad_norm": 7.495645523071289,
       "learning_rate": 1e-06,
-      "loss": 2.268,
+      "loss": 2.4391,
       "step": 55
     },
     {
       "epoch": 0.8333333333333334,
-      "grad_norm": 2.464299201965332,
+      "grad_norm": 5.458991527557373,
       "learning_rate": 1e-06,
-      "loss": 2.45,
+      "loss": 2.2362,
       "step": 60
     },
     {
       "epoch": 0.9027777777777778,
-      "grad_norm": 2.9303369522094727,
+      "grad_norm": 5.659170627593994,
       "learning_rate": 1e-06,
-      "loss": 2.5548,
+      "loss": 2.2188,
       "step": 65
     },
     {
       "epoch": 0.9722222222222222,
-      "grad_norm": 1.658677339553833,
+      "grad_norm": 4.8217997550964355,
       "learning_rate": 1e-06,
-      "loss": 2.3566,
+      "loss": 2.154,
       "step": 70
     },
     {
-      "epoch": 1.0,
-      "step": 72,
-      "total_flos": 4.988916731582874e+16,
-      "train_loss": 2.5252017312579684,
-      "train_runtime": 672.6439,
-      "train_samples_per_second": 1.713,
-      "train_steps_per_second": 0.107
+      "epoch": 1.0416666666666667,
+      "grad_norm": 6.693627834320068,
+      "learning_rate": 1e-06,
+      "loss": 2.1182,
+      "step": 75
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 6.67255973815918,
+      "learning_rate": 1e-06,
+      "loss": 2.1879,
+      "step": 80
+    },
+    {
+      "epoch": 1.1805555555555556,
+      "grad_norm": 4.831326007843018,
+      "learning_rate": 1e-06,
+      "loss": 2.1077,
+      "step": 85
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 4.830414295196533,
+      "learning_rate": 1e-06,
+      "loss": 2.026,
+      "step": 90
+    },
+    {
+      "epoch": 1.3194444444444444,
+      "grad_norm": 5.039080619812012,
+      "learning_rate": 1e-06,
+      "loss": 2.0585,
+      "step": 95
+    },
+    {
+      "epoch": 1.3888888888888888,
+      "grad_norm": 5.749475002288818,
+      "learning_rate": 1e-06,
+      "loss": 2.0486,
+      "step": 100
+    },
+    {
+      "epoch": 1.4583333333333333,
+      "grad_norm": 5.0571770668029785,
+      "learning_rate": 1e-06,
+      "loss": 1.9616,
+      "step": 105
+    },
+    {
+      "epoch": 1.5277777777777777,
+      "grad_norm": 4.597809314727783,
+      "learning_rate": 1e-06,
+      "loss": 1.9063,
+      "step": 110
+    },
+    {
+      "epoch": 1.5972222222222223,
+      "grad_norm": 5.453224182128906,
+      "learning_rate": 1e-06,
+      "loss": 2.1802,
+      "step": 115
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 4.519564628601074,
+      "learning_rate": 1e-06,
+      "loss": 2.2049,
+      "step": 120
+    },
+    {
+      "epoch": 1.7361111111111112,
+      "grad_norm": 4.976806163787842,
+      "learning_rate": 1e-06,
+      "loss": 1.8881,
+      "step": 125
+    },
+    {
+      "epoch": 1.8055555555555556,
+      "grad_norm": 4.543058395385742,
+      "learning_rate": 1e-06,
+      "loss": 2.0673,
+      "step": 130
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 4.89597225189209,
+      "learning_rate": 1e-06,
+      "loss": 2.0158,
+      "step": 135
+    },
+    {
+      "epoch": 1.9444444444444444,
+      "grad_norm": 4.95186185836792,
+      "learning_rate": 1e-06,
+      "loss": 2.0052,
+      "step": 140
+    },
+    {
+      "epoch": 2.0,
+      "step": 144,
+      "total_flos": 9.601343958286336e+16,
+      "train_loss": 2.2213990655210285,
+      "train_runtime": 2501.8045,
+      "train_samples_per_second": 0.921,
+      "train_steps_per_second": 0.058
     }
   ],
   "logging_steps": 5,
-  "max_steps": 72,
+  "max_steps": 144,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 2,
   "save_steps": 200,
-  "stateful_callbacks": {
-    "TrainerControl": {
-      "args": {
-        "should_epoch_stop": false,
-        "should_evaluate": false,
-        "should_log": false,
-        "should_save": true,
-        "should_training_stop": true
-      },
-      "attributes": {}
-    }
-  },
-  "total_flos": 4.988916731582874e+16,
-  "train_batch_size": 8,
+  "total_flos": 9.601343958286336e+16,
+  "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:95877359ada36ac411af5909de5cef10f22c8118e25725b3ac062bf5aec92941
-size 6968
+oid sha256:4bce47b54376c8018ec37390823c002788eed73a220893b64e1cf33e79ab99a9
+size 6776