Add files using upload-large-folder tool
Browse files- checkpoints/epoch_1.pt +3 -0
- checkpoints/epoch_10.pt +3 -0
- checkpoints/epoch_11.pt +3 -0
- checkpoints/epoch_12.pt +3 -0
- checkpoints/epoch_13.pt +3 -0
- checkpoints/epoch_14.pt +3 -0
- checkpoints/epoch_15.pt +3 -0
- checkpoints/epoch_16.pt +3 -0
- checkpoints/epoch_17.pt +3 -0
- checkpoints/epoch_18.pt +3 -0
- checkpoints/epoch_19.pt +3 -0
- checkpoints/epoch_2.pt +3 -0
- checkpoints/epoch_20.pt +3 -0
- checkpoints/epoch_21.pt +3 -0
- checkpoints/epoch_22.pt +3 -0
- checkpoints/epoch_23.pt +3 -0
- checkpoints/epoch_24.pt +3 -0
- checkpoints/epoch_25.pt +3 -0
- checkpoints/epoch_26.pt +3 -0
- checkpoints/epoch_27.pt +3 -0
- checkpoints/epoch_28.pt +3 -0
- checkpoints/epoch_29.pt +3 -0
- checkpoints/epoch_3.pt +3 -0
- checkpoints/epoch_30.pt +3 -0
- checkpoints/epoch_31.pt +3 -0
- checkpoints/epoch_32.pt +3 -0
- checkpoints/epoch_4.pt +3 -0
- checkpoints/epoch_5.pt +3 -0
- checkpoints/epoch_6.pt +3 -0
- checkpoints/epoch_7.pt +3 -0
- checkpoints/epoch_8.pt +3 -0
- checkpoints/epoch_9.pt +3 -0
- checkpoints/results.jsonl +16 -0
- config.json +46 -0
- merges.txt +0 -0
- model.safetensors +3 -0
- out.log +0 -0
- params.txt +103 -0
- preprocessor_config.json +28 -0
- special_tokens_map.json +30 -0
- tokenizer_config.json +33 -0
- vocab.json +0 -0
checkpoints/epoch_1.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d4348ed21b0199994e889c1443c72594e61dc73a10635bfc50a26b1b6e00d0d5
|
| 3 |
+
size 560073734
|
checkpoints/epoch_10.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:669a96da0c5032592129c7b1028104a7a1e28ba798991defbd436a6eb3cae14e
|
| 3 |
+
size 560075406
|
checkpoints/epoch_11.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dec7d2dad5dde5fdda3b8ca39931690f121d93d28d2df5e058486fdeec2ae420
|
| 3 |
+
size 560075406
|
checkpoints/epoch_12.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:824544010c35ec74d4922d44cc54d4558631e1ea191f290cbd7e932bf3222985
|
| 3 |
+
size 560075406
|
checkpoints/epoch_13.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:83a476fb714176a6ee2506594a37bc8d2a090868cc14c5c995e7f28092fbb689
|
| 3 |
+
size 560075406
|
checkpoints/epoch_14.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:118d6339ee298db69a81a8fc53843b566c29669779aa96751cc0df6b225abd19
|
| 3 |
+
size 560075406
|
checkpoints/epoch_15.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4147a1a1bed3bbcbaa429006a44026e58171d3dc4d5dcbad8275d874a5b45a59
|
| 3 |
+
size 560075406
|
checkpoints/epoch_16.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7d20bf146a936dd0d49ad5d273e1ef8b4bc3be204b4b6a3c16e21dd87aef096f
|
| 3 |
+
size 560075406
|
checkpoints/epoch_17.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:43e8d438ff3734c3439fda00056f06d260e8c06baa24c9a1497204db667f8310
|
| 3 |
+
size 560075406
|
checkpoints/epoch_18.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:19513bd5824d4b6973748c3b9fc12dfd850da3fa71a995e73ad56c28d91ba7ac
|
| 3 |
+
size 560075406
|
checkpoints/epoch_19.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8f041f91aeaf066ab8c9250fa3c319d0bd5a23fed7f5c6f1bf2b5324965ec525
|
| 3 |
+
size 560075406
|
checkpoints/epoch_2.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9a8ce97b86e32fa375be55ae738655f67b24905ea958cad5fab1cd4b1860f6ee
|
| 3 |
+
size 560073734
|
checkpoints/epoch_20.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f1c894bf535fb1ea82a564674e666d7c4608c748f579d43947d5cdb51ae93184
|
| 3 |
+
size 560075406
|
checkpoints/epoch_21.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:00cec8a7a9e136015c8c0d9e9d5cfb828b6ee370483281802c23b637bd3e99aa
|
| 3 |
+
size 560075406
|
checkpoints/epoch_22.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:43e50a0005df9711f488d10d91f8a834551ba77b01cf33aef9c880a7d3db781f
|
| 3 |
+
size 560075406
|
checkpoints/epoch_23.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eaf0263e76020b8d9eb93f8f09accf0540cd395dd2c7ad8e4dfb4c743e11fd15
|
| 3 |
+
size 560075406
|
checkpoints/epoch_24.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:266e7ba69d5db186bb8f01c9e76f8b038b09fd26e626211a45e7b10578963855
|
| 3 |
+
size 560075406
|
checkpoints/epoch_25.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:90398df9a2ec91b77e2b806d1a95ed6320250f02f1c73759ce256b6ea1be0bfc
|
| 3 |
+
size 560075406
|
checkpoints/epoch_26.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d8ad649677e2819c6264a56d2aa8b64bd230d740cd26e6ae158b02c36ba024ca
|
| 3 |
+
size 560075406
|
checkpoints/epoch_27.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e12dba4a2dfbe1485ff79beed6f0c83b4e544ad18a911eebc9ce408f8798e15a
|
| 3 |
+
size 560075406
|
checkpoints/epoch_28.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df4c887e1d72442fb59e6792860df8749bd5cb728ee0425bb672e2451fe55d5d
|
| 3 |
+
size 560075406
|
checkpoints/epoch_29.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8789def418293864579c71d1c61a28ba422d1981a3e0153b435679a9ece04e64
|
| 3 |
+
size 560075406
|
checkpoints/epoch_3.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc6a99a91a6f830e7d1b738f3080bfe9efffcacf2d03cb9dece426baef78cb7b
|
| 3 |
+
size 560073734
|
checkpoints/epoch_30.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3933beb9dc087e1e8b7d8e7c4397a698aa87dac5e93ee350b45ce067754a47d3
|
| 3 |
+
size 560075406
|
checkpoints/epoch_31.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:928a3123003c8cca3735d2e41adea8dd89cfcbdc1ef27f54e1f3199019d5ad1e
|
| 3 |
+
size 560075406
|
checkpoints/epoch_32.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:84028b88b1081f534db4642ee24e99acb014f3977592717dc867f41b0fa082eb
|
| 3 |
+
size 560075406
|
checkpoints/epoch_4.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:668e8261bfe910b6ed87b85bfa5785d4d3beac208b20957985b37840f5ab1b93
|
| 3 |
+
size 560073734
|
checkpoints/epoch_5.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e26980a371669db9c378ece120a9047c5fe8b530c2715cd870b874e99fef9227
|
| 3 |
+
size 560073734
|
checkpoints/epoch_6.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9687df30c13c8328650b3398beee454e54496b7b326615f987e6c4bba7706236
|
| 3 |
+
size 560073734
|
checkpoints/epoch_7.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5208b812efede45110a383d6b59189cfbaa5d9333d291e74dc7b2b60c2df3ea7
|
| 3 |
+
size 560073734
|
checkpoints/epoch_8.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8aba78335ee3694c50c5e929ca2d9d97dac27d61f55ab34ad1a79f7d59137518
|
| 3 |
+
size 560073734
|
checkpoints/epoch_9.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b4fc2554ef97fa679838083dce5395f2e040b85f4f5a68348ad20f4df636b594
|
| 3 |
+
size 560073734
|
checkpoints/results.jsonl
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"imagenet-zeroshot-val-top1": 0.10186, "imagenet-zeroshot-val-top5": 0.25288}
|
| 2 |
+
{"imagenet-zeroshot-val-top1": 0.16618, "imagenet-zeroshot-val-top5": 0.37234}
|
| 3 |
+
{"imagenet-zeroshot-val-top1": 0.18072, "imagenet-zeroshot-val-top5": 0.3958}
|
| 4 |
+
{"imagenet-zeroshot-val-top1": 0.20992, "imagenet-zeroshot-val-top5": 0.44426}
|
| 5 |
+
{"imagenet-zeroshot-val-top1": 0.22744, "imagenet-zeroshot-val-top5": 0.46702}
|
| 6 |
+
{"imagenet-zeroshot-val-top1": 0.23398, "imagenet-zeroshot-val-top5": 0.47998}
|
| 7 |
+
{"imagenet-zeroshot-val-top1": 0.25294, "imagenet-zeroshot-val-top5": 0.50992}
|
| 8 |
+
{"imagenet-zeroshot-val-top1": 0.25416, "imagenet-zeroshot-val-top5": 0.50892}
|
| 9 |
+
{"imagenet-zeroshot-val-top1": 0.26532, "imagenet-zeroshot-val-top5": 0.52794}
|
| 10 |
+
{"imagenet-zeroshot-val-top1": 0.26664, "imagenet-zeroshot-val-top5": 0.52882}
|
| 11 |
+
{"imagenet-zeroshot-val-top1": 0.2744, "imagenet-zeroshot-val-top5": 0.53928}
|
| 12 |
+
{"imagenet-zeroshot-val-top1": 0.28536, "imagenet-zeroshot-val-top5": 0.54932}
|
| 13 |
+
{"imagenet-zeroshot-val-top1": 0.29048, "imagenet-zeroshot-val-top5": 0.55966}
|
| 14 |
+
{"imagenet-zeroshot-val-top1": 0.29438, "imagenet-zeroshot-val-top5": 0.5631}
|
| 15 |
+
{"imagenet-zeroshot-val-top1": 0.29654, "imagenet-zeroshot-val-top5": 0.56632}
|
| 16 |
+
{"imagenet-zeroshot-val-top1": 0.29696, "imagenet-zeroshot-val-top5": 0.56852}
|
config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"CLIPModel"
|
| 4 |
+
],
|
| 5 |
+
"initializer_factor": 1.0,
|
| 6 |
+
"logit_scale_init_value": 2.6592,
|
| 7 |
+
"model_type": "clip",
|
| 8 |
+
"projection_dim": 512,
|
| 9 |
+
"text_config": {
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"bos_token_id": 0,
|
| 12 |
+
"dropout": 0.0,
|
| 13 |
+
"eos_token_id": 2,
|
| 14 |
+
"hidden_act": "quick_gelu",
|
| 15 |
+
"hidden_size": 384,
|
| 16 |
+
"initializer_factor": 1.0,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 1536,
|
| 19 |
+
"layer_norm_eps": 1e-05,
|
| 20 |
+
"max_position_embeddings": 77,
|
| 21 |
+
"model_type": "clip_text_model",
|
| 22 |
+
"num_attention_heads": 6,
|
| 23 |
+
"num_hidden_layers": 12,
|
| 24 |
+
"projection_dim": 512,
|
| 25 |
+
"vocab_size": 49408
|
| 26 |
+
},
|
| 27 |
+
"torch_dtype": "float32",
|
| 28 |
+
"transformers_version": "4.53.0",
|
| 29 |
+
"vision_config": {
|
| 30 |
+
"attention_dropout": 0.0,
|
| 31 |
+
"dropout": 0.0,
|
| 32 |
+
"hidden_act": "quick_gelu",
|
| 33 |
+
"hidden_size": 192,
|
| 34 |
+
"image_size": 224,
|
| 35 |
+
"initializer_factor": 1.0,
|
| 36 |
+
"initializer_range": 0.02,
|
| 37 |
+
"intermediate_size": 768,
|
| 38 |
+
"layer_norm_eps": 1e-05,
|
| 39 |
+
"model_type": "clip_vision_model",
|
| 40 |
+
"num_attention_heads": 3,
|
| 41 |
+
"num_channels": 3,
|
| 42 |
+
"num_hidden_layers": 12,
|
| 43 |
+
"patch_size": 16,
|
| 44 |
+
"projection_dim": 5
|
| 45 |
+
}
|
| 46 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4b4dafe19e3f186f7ed27516d43b5a42c3b0ade3dc586f4ca832cecb7bdad430
|
| 3 |
+
size 184511660
|
out.log
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
params.txt
ADDED
|
@@ -0,0 +1,103 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
accum_freq: 1
|
| 2 |
+
aug_cfg: {}
|
| 3 |
+
batch_size: 256
|
| 4 |
+
beta1: 0.9
|
| 5 |
+
beta2: 0.98
|
| 6 |
+
cache_dir: None
|
| 7 |
+
checkpoint_path: ./models/ViT-T-16-CC9M-distill-DFN2B-CLIP-ViT-L-14-MSE/checkpoints
|
| 8 |
+
coca_caption_loss_weight: 2.0
|
| 9 |
+
coca_contrastive_loss_weight: 1.0
|
| 10 |
+
copy_codebase: False
|
| 11 |
+
csv_caption_key: text
|
| 12 |
+
csv_img_key: image
|
| 13 |
+
csv_separator:
|
| 14 |
+
dataset_resampled: False
|
| 15 |
+
dataset_type: auto
|
| 16 |
+
ddp_static_graph: False
|
| 17 |
+
debug: False
|
| 18 |
+
delete_previous_checkpoint: False
|
| 19 |
+
device: cuda:0
|
| 20 |
+
dist_backend: None
|
| 21 |
+
dist_url: None
|
| 22 |
+
distill: True
|
| 23 |
+
distill_model: local-dir:/ist-project/scads/pumet/models/DFN2B-CLIP-ViT-L-14
|
| 24 |
+
distill_pretrained: None
|
| 25 |
+
distributed: True
|
| 26 |
+
epochs: 32
|
| 27 |
+
epochs_cooldown: None
|
| 28 |
+
eps: 1e-06
|
| 29 |
+
force_context_length: None
|
| 30 |
+
force_custom_text: False
|
| 31 |
+
force_image_size: None
|
| 32 |
+
force_patch_dropout: None
|
| 33 |
+
force_quick_gelu: False
|
| 34 |
+
gather_with_grad: False
|
| 35 |
+
grad_checkpointing: False
|
| 36 |
+
grad_clip_norm: None
|
| 37 |
+
horovod: False
|
| 38 |
+
image_interpolation: None
|
| 39 |
+
image_mean: None
|
| 40 |
+
image_resize_mode: None
|
| 41 |
+
image_std: None
|
| 42 |
+
imagenet_v2: None
|
| 43 |
+
imagenet_val: /ist-project/scads/pumet/datasets/imagenet-1k/validation
|
| 44 |
+
local_loss: False
|
| 45 |
+
local_rank: 0
|
| 46 |
+
lock_image: False
|
| 47 |
+
lock_image_freeze_bn_stats: False
|
| 48 |
+
lock_image_unlocked_groups: 0
|
| 49 |
+
lock_text: False
|
| 50 |
+
lock_text_freeze_layer_norm: False
|
| 51 |
+
lock_text_unlocked_layers: 0
|
| 52 |
+
log_every_n_steps: 100
|
| 53 |
+
log_level: 20
|
| 54 |
+
log_local: False
|
| 55 |
+
log_path: ./models/ViT-T-16-CC9M-distill-DFN2B-CLIP-ViT-L-14-MSE/out.log
|
| 56 |
+
logs: ./models/
|
| 57 |
+
loss_dist_impl: None
|
| 58 |
+
lr: 0.0005
|
| 59 |
+
lr_cooldown_end: 0.0
|
| 60 |
+
lr_cooldown_power: 1.0
|
| 61 |
+
lr_scheduler: cosine
|
| 62 |
+
model: ViT-T-16
|
| 63 |
+
momentum: None
|
| 64 |
+
name: ViT-T-16-CC9M-distill-DFN2B-CLIP-ViT-L-14-MSE
|
| 65 |
+
no_set_device_rank: False
|
| 66 |
+
opt: adamw
|
| 67 |
+
precision: amp
|
| 68 |
+
pretrained:
|
| 69 |
+
pretrained_image: False
|
| 70 |
+
rank: 0
|
| 71 |
+
remote_sync: None
|
| 72 |
+
remote_sync_frequency: 300
|
| 73 |
+
remote_sync_protocol: s3
|
| 74 |
+
report_to: none
|
| 75 |
+
resume: None
|
| 76 |
+
s_embed: 512
|
| 77 |
+
save_frequency: 1
|
| 78 |
+
save_most_recent: False
|
| 79 |
+
seed: 0
|
| 80 |
+
siglip: False
|
| 81 |
+
skip_scheduler: False
|
| 82 |
+
t_embed: 768
|
| 83 |
+
tensorboard: False
|
| 84 |
+
tensorboard_path:
|
| 85 |
+
torchcompile: False
|
| 86 |
+
torchscript: False
|
| 87 |
+
trace: False
|
| 88 |
+
train_data: /ist-project/scads/pumet/datasets/cc9m/cc9m.csv
|
| 89 |
+
train_data_upsampling_factors: None
|
| 90 |
+
train_num_samples: None
|
| 91 |
+
use_bn_sync: False
|
| 92 |
+
use_bnb_linear: None
|
| 93 |
+
val_data: None
|
| 94 |
+
val_frequency: 1
|
| 95 |
+
val_num_samples: None
|
| 96 |
+
wandb: False
|
| 97 |
+
wandb_notes:
|
| 98 |
+
wandb_project_name: open-clip
|
| 99 |
+
warmup: 2000
|
| 100 |
+
wd: 0.2
|
| 101 |
+
workers: 8
|
| 102 |
+
world_size: 16
|
| 103 |
+
zeroshot_frequency: 2
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"crop_size": {
|
| 3 |
+
"height": 224,
|
| 4 |
+
"width": 224
|
| 5 |
+
},
|
| 6 |
+
"do_center_crop": true,
|
| 7 |
+
"do_convert_rgb": true,
|
| 8 |
+
"do_normalize": true,
|
| 9 |
+
"do_rescale": true,
|
| 10 |
+
"do_resize": true,
|
| 11 |
+
"image_mean": [
|
| 12 |
+
0.48145466,
|
| 13 |
+
0.4578275,
|
| 14 |
+
0.40821073
|
| 15 |
+
],
|
| 16 |
+
"image_processor_type": "CLIPImageProcessor",
|
| 17 |
+
"image_std": [
|
| 18 |
+
0.26862954,
|
| 19 |
+
0.26130258,
|
| 20 |
+
0.27577711
|
| 21 |
+
],
|
| 22 |
+
"processor_class": "CLIPProcessor",
|
| 23 |
+
"resample": 3,
|
| 24 |
+
"rescale_factor": 0.00392156862745098,
|
| 25 |
+
"size": {
|
| 26 |
+
"shortest_edge": 224
|
| 27 |
+
}
|
| 28 |
+
}
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<|startoftext|>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": true,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "<|endoftext|>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "<|endoftext|>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "<|endoftext|>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"49406": {
|
| 5 |
+
"content": "<|startoftext|>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": true,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"49407": {
|
| 13 |
+
"content": "<|endoftext|>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": false,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
}
|
| 20 |
+
},
|
| 21 |
+
"bos_token": "<|startoftext|>",
|
| 22 |
+
"clean_up_tokenization_spaces": false,
|
| 23 |
+
"do_lower_case": true,
|
| 24 |
+
"eos_token": "<|endoftext|>",
|
| 25 |
+
"errors": "replace",
|
| 26 |
+
"extra_special_tokens": {},
|
| 27 |
+
"model_max_length": 77,
|
| 28 |
+
"pad_token": "<|endoftext|>",
|
| 29 |
+
"processor_class": "CLIPProcessor",
|
| 30 |
+
"tokenizer_class": "CLIPTokenizer",
|
| 31 |
+
"unk_token": "<|endoftext|>",
|
| 32 |
+
"use_fast": false
|
| 33 |
+
}
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|