trocr-invoice

README.md

@@ -14,8 +14,13 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [microsoft/trocr-base-stage1](https://huggingface.co/microsoft/trocr-base-stage1) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4098
-- Cer: 0.0324
+- eval_loss: 1.7428
+- eval_cer: 0.1176
+- eval_runtime: 578.8949
+- eval_samples_per_second: 7.692
+- eval_steps_per_second: 0.962
+- epoch: 2.0
+- step: 8906
 
 ## Model description
 
@@ -40,25 +45,9 @@ The following hyperparameters were used during training:
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 10
+- num_epochs: 3
 - mixed_precision_training: Native AMP
 
-### Training results
-
-| Training Loss | Epoch | Step | Validation Loss | Cer    |
-|:-------------:|:-----:|:----:|:---------------:|:------:|
-| No log        | 1.0   | 390  | 0.5628          | 0.1036 |
-| 0.9189        | 2.0   | 780  | 0.4374          | 0.0472 |
-| 0.4488        | 3.0   | 1170 | 0.4076          | 0.0300 |
-| 0.378         | 4.0   | 1560 | 0.3955          | 0.0239 |
-| 0.378         | 5.0   | 1950 | 0.3553          | 0.0150 |
-| 0.336         | 6.0   | 2340 | 0.3574          | 0.0132 |
-| 0.3207        | 7.0   | 2730 | 0.3441          | 0.0103 |
-| 0.3079        | 8.0   | 3120 | 0.3473          | 0.0121 |
-| 0.26          | 9.0   | 3510 | 0.3490          | 0.0121 |
-| 0.26          | 10.0  | 3900 | 0.3513          | 0.0113 |
-
-
 ### Framework versions
 
 - Transformers 4.42.4

config.json

@@ -165,7 +165,7 @@
   "eos_token_id": 2,
   "is_encoder_decoder": true,
   "length_penalty": 2.0,
-  "max_length": 64,
+  "max_length": 128,
   "model_type": "vision-encoder-decoder",
   "no_repeat_ngram_size": 3,
   "num_beams": 4,

generation_config.json

@@ -4,7 +4,7 @@
   "early_stopping": true,
   "eos_token_id": 2,
   "length_penalty": 2.0,
-  "max_length": 64,
+  "max_length": 128,
   "no_repeat_ngram_size": 3,
   "num_beams": 4,
   "pad_token_id": 1,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:749514458b6c6e60e2e7d9682a44e51d4bc0e8d4747718b9297b016e838eb5b8
+oid sha256:ecc40fd68bb1f9dd2369f6c9345c46a6f2a066663781d7640a5753e340aa05b7
 size 1539518812

runs/Jul20_14-29-49_201cff9e7870/events.out.tfevents.1721485800.201cff9e7870.898.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29ff7c640ff29b043732bcd9791978eaa700f3395b7ee00e8186402ed96235a8
+size 9130

runs/Jul20_14-33-52_201cff9e7870/events.out.tfevents.1721486033.201cff9e7870.898.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f19e4b13b4d02e151f09925cc4bcfffd8635912762e1c4e076fb839016cad5fb
+size 8280

runs/Jul20_14-35-19_201cff9e7870/events.out.tfevents.1721486120.201cff9e7870.9490.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9b2702c55e81fad12615947b9e7abc521d8d9e96d03b11a7bcc8f047ad6e31f
+size 13775

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6d361a886d0b23c9cae2e40a342f7001ef8801a281aaf2d29d0249bb45ac5521
+oid sha256:62ed26fe8d87087b3c7cdaa09993b1a6c804d0ac6d7afd2b96ed42809ca8bf4a
 size 5304