Training in progress, epoch 1
Browse files
config.json
CHANGED
|
@@ -40,7 +40,7 @@
|
|
| 40 |
"use_cache": true,
|
| 41 |
"vocab_size": 50304
|
| 42 |
},
|
| 43 |
-
"decoder_start_token_id":
|
| 44 |
"dtype": "float32",
|
| 45 |
"encoder": {
|
| 46 |
"_name_or_path": "microsoft/dit-base",
|
|
@@ -58,9 +58,10 @@
|
|
| 58 |
"hidden_act": "gelu",
|
| 59 |
"hidden_dropout_prob": 0.0,
|
| 60 |
"hidden_size": 768,
|
| 61 |
-
"image_size":
|
| 62 |
"initializer_range": 0.02,
|
| 63 |
"intermediate_size": 3072,
|
|
|
|
| 64 |
"layer_norm_eps": 1e-12,
|
| 65 |
"layer_scale_init_value": 0.1,
|
| 66 |
"model_type": "beit",
|
|
@@ -116,7 +117,7 @@
|
|
| 116 |
"max_length": null,
|
| 117 |
"model_type": "vision-encoder-decoder",
|
| 118 |
"no_repeat_ngram_size": null,
|
| 119 |
-
"pad_token_id":
|
| 120 |
"tie_word_embeddings": false,
|
| 121 |
"transformers_version": "4.57.6"
|
| 122 |
}
|
|
|
|
| 40 |
"use_cache": true,
|
| 41 |
"vocab_size": 50304
|
| 42 |
},
|
| 43 |
+
"decoder_start_token_id": 50292,
|
| 44 |
"dtype": "float32",
|
| 45 |
"encoder": {
|
| 46 |
"_name_or_path": "microsoft/dit-base",
|
|
|
|
| 58 |
"hidden_act": "gelu",
|
| 59 |
"hidden_dropout_prob": 0.0,
|
| 60 |
"hidden_size": 768,
|
| 61 |
+
"image_size": 1024,
|
| 62 |
"initializer_range": 0.02,
|
| 63 |
"intermediate_size": 3072,
|
| 64 |
+
"interpolate_pos_encoding": true,
|
| 65 |
"layer_norm_eps": 1e-12,
|
| 66 |
"layer_scale_init_value": 0.1,
|
| 67 |
"model_type": "beit",
|
|
|
|
| 117 |
"max_length": null,
|
| 118 |
"model_type": "vision-encoder-decoder",
|
| 119 |
"no_repeat_ngram_size": null,
|
| 120 |
+
"pad_token_id": 50270,
|
| 121 |
"tie_word_embeddings": false,
|
| 122 |
"transformers_version": "4.57.6"
|
| 123 |
}
|
generation_config.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
"_from_model_config": true,
|
| 3 |
"bos_token_id": 50256,
|
| 4 |
-
"decoder_start_token_id":
|
| 5 |
"eos_token_id": 50256,
|
| 6 |
"max_length": 768,
|
| 7 |
"no_repeat_ngram_size": 3,
|
| 8 |
-
"pad_token_id":
|
| 9 |
"transformers_version": "4.57.6"
|
| 10 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"_from_model_config": true,
|
| 3 |
"bos_token_id": 50256,
|
| 4 |
+
"decoder_start_token_id": 50292,
|
| 5 |
"eos_token_id": 50256,
|
| 6 |
"max_length": 768,
|
| 7 |
"no_repeat_ngram_size": 3,
|
| 8 |
+
"pad_token_id": 50270,
|
| 9 |
"transformers_version": "4.57.6"
|
| 10 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dfe6249d11963ce213ad4381c01231ccf4bedf5833f33387364bbd3cc2683c24
|
| 3 |
+
size 954658376
|
runs/Jan24_19-48-57_a4dbac9104bf/events.out.tfevents.1769284170.a4dbac9104bf.6059.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e221ea8586865580bbff389fcb27676e82106cdb7f3d8c09d8c4d2d51f6b948f
|
| 3 |
+
size 7509
|
runs/Jan24_19-49-57_a4dbac9104bf/events.out.tfevents.1769284199.a4dbac9104bf.6059.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5df0d60b7b5c0d4d53641aea9447ce7f481e242a275cfda6dbce2ddf4c5312ff
|
| 3 |
+
size 10690
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8370a4dccc97227cb977d535f9a8edf10298c6f4038c99258853615144c17db
|
| 3 |
size 5905
|