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",
|
|
@@ -58,7 +58,7 @@
|
|
| 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 |
"interpolate_pos_encoding": true,
|
|
@@ -117,7 +117,7 @@
|
|
| 117 |
"max_length": null,
|
| 118 |
"model_type": "vision-encoder-decoder",
|
| 119 |
"no_repeat_ngram_size": null,
|
| 120 |
-
"pad_token_id":
|
| 121 |
"tie_word_embeddings": false,
|
| 122 |
"transformers_version": "4.57.6"
|
| 123 |
}
|
|
|
|
| 40 |
"use_cache": true,
|
| 41 |
"vocab_size": 50304
|
| 42 |
},
|
| 43 |
+
"decoder_start_token_id": 50287,
|
| 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": 2048,
|
| 62 |
"initializer_range": 0.02,
|
| 63 |
"intermediate_size": 3072,
|
| 64 |
"interpolate_pos_encoding": true,
|
|
|
|
| 117 |
"max_length": null,
|
| 118 |
"model_type": "vision-encoder-decoder",
|
| 119 |
"no_repeat_ngram_size": null,
|
| 120 |
+
"pad_token_id": 50257,
|
| 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": 50287,
|
| 5 |
"eos_token_id": 50256,
|
| 6 |
"max_length": 768,
|
| 7 |
"no_repeat_ngram_size": 3,
|
| 8 |
+
"pad_token_id": 50257,
|
| 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 954658376
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c281d1f01a17f4e67a565449eb0094aa497a75c2c43836005d09198e185e62b9
|
| 3 |
size 954658376
|
runs/Jan24_19-49-57_a4dbac9104bf/events.out.tfevents.1769284199.a4dbac9104bf.6059.3
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 sha256:568f6c629bd2075846a93e4a5adcdb6490da2207e36fdb55b4b9df2b1479402f
|
| 3 |
+
size 55780
|
runs/Jan24_21-14-52_a4dbac9104bf/events.out.tfevents.1769289295.a4dbac9104bf.22724.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6bb7714126a44b26cc83b8f2aeef263e5fdd8cf4e7c7ad64e8b7016d00af6b63
|
| 3 |
+
size 13855
|
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 sha256:07f35543c4fb9489c96b0dc976b6eacd83dcb758339dffde476bab1a5f636bd5
|
| 3 |
size 5905
|