Training checkpoint

README.md

@@ -1,22 +1,24 @@
 ---
-base_model: openai/whisper-small
 library_name: transformers
+language:
+- en
 license: apache-2.0
-metrics:
-- wer
+base_model: openai/whisper-small
 tags:
 - generated_from_trainer
+metrics:
+- wer
 model-index:
-- name: Whisper-squeezeformer-v3
+- name: Whisper-squeezeformer-NSQU-whisper
   results: []
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 should probably proofread and complete it, then remove this comment. -->
 
-# Whisper-squeezeformer-v3
+# Whisper-squeezeformer-NSQU-whisper
 
-This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the None dataset.
+This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the LibriSpeech dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.1297
 - Wer: 5.6204

trainer_state.json

@@ -1,9 +1,9 @@
 {
-  "best_metric": 6.75973828362751,
-  "best_model_checkpoint": "./Whisper-squeezeformer-v3\\checkpoint-27500",
-  "epoch": 12.0,
+  "best_metric": 5.599513085818624,
+  "best_model_checkpoint": "./Whisper-squeezeformer-v3\\checkpoint-40000",
+  "epoch": 18.0,
   "eval_steps": 2500,
-  "global_step": 30000,
+  "global_step": 45000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -201,19 +201,115 @@
       "step": 30000
     },
     {
-      "epoch": 12.0,
-      "step": 30000,
-      "total_flos": 2.07987720192e+20,
-      "train_loss": 0.04717991434733073,
-      "train_runtime": 48062.4633,
-      "train_samples_per_second": 12.484,
-      "train_steps_per_second": 0.624
+      "epoch": 13.0,
+      "grad_norm": 4.114705562591553,
+      "learning_rate": 2.004e-06,
+      "loss": 0.1946,
+      "step": 32500
+    },
+    {
+      "epoch": 13.0,
+      "eval_loss": 0.139073446393013,
+      "eval_runtime": 595.1979,
+      "eval_samples_per_second": 4.402,
+      "eval_steps_per_second": 0.551,
+      "eval_wer": 6.421180766889837,
+      "step": 32500
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 4.109494686126709,
+      "learning_rate": 1.3378666666666667e-06,
+      "loss": 0.1425,
+      "step": 35000
+    },
+    {
+      "epoch": 14.0,
+      "eval_loss": 0.13689054548740387,
+      "eval_runtime": 633.0271,
+      "eval_samples_per_second": 4.139,
+      "eval_steps_per_second": 0.518,
+      "eval_wer": 5.87530432136336,
+      "step": 35000
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 2.726956367492676,
+      "learning_rate": 6.712e-07,
+      "loss": 0.1145,
+      "step": 37500
+    },
+    {
+      "epoch": 15.0,
+      "eval_loss": 0.13682714104652405,
+      "eval_runtime": 601.9865,
+      "eval_samples_per_second": 4.352,
+      "eval_steps_per_second": 0.545,
+      "eval_wer": 5.753575776019477,
+      "step": 37500
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 3.426710844039917,
+      "learning_rate": 1.1807058823529414e-06,
+      "loss": 0.1776,
+      "step": 40000
+    },
+    {
+      "epoch": 16.0,
+      "eval_loss": 0.13023081421852112,
+      "eval_runtime": 578.9166,
+      "eval_samples_per_second": 4.526,
+      "eval_steps_per_second": 0.567,
+      "eval_wer": 5.599513085818624,
+      "step": 40000
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 3.3029887676239014,
+      "learning_rate": 5.927058823529412e-07,
+      "loss": 0.1416,
+      "step": 42500
+    },
+    {
+      "epoch": 17.0,
+      "eval_loss": 0.12984110414981842,
+      "eval_runtime": 624.3393,
+      "eval_samples_per_second": 4.196,
+      "eval_steps_per_second": 0.525,
+      "eval_wer": 5.620435179549604,
+      "step": 42500
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 3.106375217437744,
+      "learning_rate": 4.705882352941177e-09,
+      "loss": 0.1239,
+      "step": 45000
+    },
+    {
+      "epoch": 18.0,
+      "eval_loss": 0.12974976003170013,
+      "eval_runtime": 580.9928,
+      "eval_samples_per_second": 4.51,
+      "eval_steps_per_second": 0.565,
+      "eval_wer": 5.620435179549604,
+      "step": 45000
+    },
+    {
+      "epoch": 18.0,
+      "step": 45000,
+      "total_flos": 3.11981580288e+20,
+      "train_loss": 0.024614295111762153,
+      "train_runtime": 36051.4679,
+      "train_samples_per_second": 24.964,
+      "train_steps_per_second": 1.248
     }
   ],
   "logging_steps": 2500,
-  "max_steps": 30000,
+  "max_steps": 45000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 12,
+  "num_train_epochs": 18,
   "save_steps": 2500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -227,7 +323,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2.07987720192e+20,
+  "total_flos": 3.11981580288e+20,
   "train_batch_size": 20,
   "trial_name": null,
   "trial_params": null