End of training

README.md

@@ -20,8 +20,8 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the zeroth_korean dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.2628
-- Cer: 10.3408
+- Loss: 0.4756
+- Cer: 16.5672
 
 ## Model description
 
@@ -47,18 +47,14 @@ The following hyperparameters were used during training:
 - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 500
-- training_steps: 5000
+- training_steps: 1000
 - mixed_precision_training: Native AMP
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss | Cer     |
 |:-------------:|:------:|:----:|:---------------:|:-------:|
-| 0.4045        | 0.7184 | 1000 | 0.4426          | 14.6710 |
-| 0.2499        | 1.4368 | 2000 | 0.3460          | 12.0642 |
-| 0.1698        | 2.1552 | 3000 | 0.2951          | 10.1798 |
-| 0.1522        | 2.8736 | 4000 | 0.2705          | 10.2662 |
-| 0.1306        | 3.5920 | 5000 | 0.2628          | 10.3408 |
+| 0.4325        | 0.7184 | 1000 | 0.4756          | 16.5672 |
 
 
 ### Framework versions

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1c95bf2cb97dfbf7f03f92955dd9eb5ed2942a8292b22026b4ca744dd8e8ea12
+oid sha256:f63bb3629b17fc02dd5f38ffe86e4c4f3870ddfb81a18d2ef70dd0b74d96adfe
 size 151061672

runs/Jul03_09-44-50_Airsound3/events.out.tfevents.1751503490.Airsound3.367215.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e79a6bb9db2db2af680ce383e54c9b614b58de9c15cb1abe8f93e45f71f4c56f
+size 15807

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d8efafc29d5d8a334dc4781196278ce8296c21b7fc3cb38b52aba4db9e47f5c0
+oid sha256:a8b6099f344a968555123884ef3d9db61689ca682633096ca4554de6f4a2dc00
 size 5905