Hugging Face's logo

michael-0acf4
/
anitag2vec

Sentence Similarity
ONNX
embedding
encoding
similarity
vector
danbooru
sakugabooru
pixiv
myanimelist
Model card Files
xet
 Community
michael-0acf4 commited on
Commit
e9e73d9
·
1 Parent(s): d263deb

feat: more checkpoints

Browse files
Files changed (23) hide show
  1. README.md +17 -7
  2. pytorch/anitag2vec_1a770feaa2bd9094_c7359727bcee4f8b_i128_e10_s60203_b300_p3841472.pth +3 -0
  3. pytorch/anitag2vec_1a770feaa2bd9094_c7359727bcee4f8b_i128_e15_s60203_b300_p3841472.pth +3 -0
  4. pytorch/anitag2vec_1a770feaa2bd9094_c7359727bcee4f8b_i128_e5_s60203_b300_p3841472.pth +3 -0
  5. pytorch/{anitag2vec_e15_s50000_p1871744.pth → anitag2vec_8ea07c7d34b64b69_c7359727bcee4f8b_i128_e10_s60203_b256_p1871744.pth} +2 -2
  6. pytorch/anitag2vec_8ea07c7d34b64b69_c7359727bcee4f8b_i128_e15_s60203_b256_p1871744.pth +3 -0
  7. pytorch/anitag2vec_8ea07c7d34b64b69_c7359727bcee4f8b_i128_e20_s60203_b256_p1871744.pth +3 -0
  8. pytorch/anitag2vec_8ea07c7d34b64b69_c7359727bcee4f8b_i128_e5_s60203_b256_p1871744.pth +3 -0
  9. pytorch/anitag2vec_a75bfbdfe75fe264_c7359727bcee4f8b_i128_e10_s60203_b256_p1871744.pth +3 -0
  10. pytorch/anitag2vec_a75bfbdfe75fe264_c7359727bcee4f8b_i128_e15_s60203_b256_p1871744.pth +3 -0
  11. pytorch/anitag2vec_a75bfbdfe75fe264_c7359727bcee4f8b_i128_e5_s60203_b256_p1871744.pth +3 -0
  12. pytorch/anitag2vec_e2ef15bc7697d18c_c7359727bcee4f8b_i128_e10_s60203_b100_p1871744.pth +3 -0
  13. pytorch/anitag2vec_e2ef15bc7697d18c_c7359727bcee4f8b_i128_e15_s60203_b100_p1871744.pth +3 -0
  14. pytorch/anitag2vec_e2ef15bc7697d18c_c7359727bcee4f8b_i128_e5_s60203_b100_p1871744.pth +3 -0
  15. pytorch/errors_1a770feaa2bd9094_c7359727bcee4f8b.json +1 -0
  16. pytorch/errors_8ea07c7d34b64b69_c7359727bcee4f8b.json +1 -0
  17. pytorch/errors_a75bfbdfe75fe264_c7359727bcee4f8b.json +1 -0
  18. pytorch/errors_e2ef15bc7697d18c_c7359727bcee4f8b.json +1 -0
  19. pytorch/setup_params_1a770feaa2bd9094_c7359727bcee4f8b.json +14 -0
  20. pytorch/{setup_params.json → setup_params_8ea07c7d34b64b69_c7359727bcee4f8b.json} +5 -3
  21. pytorch/setup_params_a75bfbdfe75fe264_c7359727bcee4f8b.json +14 -0
  22. pytorch/setup_params_e2ef15bc7697d18c_c7359727bcee4f8b.json +14 -0
  23. pytorch/{token_vocab_size_5000_freq_3.json → token_dataset_c7359727bcee4f8b_vocab_size_5000_freq_3.json} +0 -0
README.md CHANGED
@@ -9,16 +9,26 @@ Training and inference examples are all available on [my github](https://github.
9
  Implementation is detailed in [this blog post](https://blog.afmichael.dev/posts/2026/set-embeddings-and-anitag2vec/).
10
 
11
  ```python
 
 
 
 
12
  device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
13
- cfg = SetupConfig.load_from_file("./pytorch/setup_params.json")
14
- data, tagtok, anitag2vec = get_setup(
15
- cfg,
16
- device=device,
17
- prefix_path= "."
 
 
 
 
 
 
 
18
  )
19
- # Load model
20
- anitag2vec.load_state_dict(torch.load("./pytorch/anitag2vec_e15_s50000_p1871744.pth"))
21
  anitag2vec.to(device)
 
22
  anitag2vec.eval()
23
  runner = AniTag2VecRunner(tagtok, anitag2vec)
24
 
 
9
  Implementation is detailed in [this blog post](https://blog.afmichael.dev/posts/2026/set-embeddings-and-anitag2vec/).
10
 
11
  ```python
12
+ TOKENIZER_PATH = "./checkpoints/token_dataset_c7359727bcee4f8b_vocab_size_5000_freq_3.json"
13
+ CONFIG_PATH = "./checkpoints/setup_params_8ea07c7d34b64b69_c7359727bcee4f8b.json"
14
+ MODEL_PATH = "./checkpoints/anitag2vec_8ea07c7d34b64b69_c7359727bcee4f8b_i128_e20_s60203_b256_p1871744.pth"
15
+
16
  device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
17
+ cfg = SetupConfig.load_from_file(CONFIG_PATH)
18
+ print(cfg)
19
+ tagtok = TagBPETokenizer(vocab_size=cfg.HYPERP_TAGTOK_VOCAB_SIZE, min_frequency=cfg.HYPERP_TAGTOK_MIN_FREQ)
20
+ tagtok.load(TOKENIZER_PATH)
21
+
22
+ anitag2vec = AniTag2Vec(
23
+ vocab_size=tagtok.vocab_size,
24
+ max_len_cut=cfg.HYPERP_TAGTOK_MAX_TOKEN_CLAMP,
25
+ d_model=cfg.HYPERP_TRANSFORMER_D_MODEL,
26
+ n_heads=cfg.HYPERP_TRANSFORMER_N_HEADS,
27
+ n_layers=cfg.HYPERP_TRANSFORMER_N_LAYERS,
28
+ output_emb=cfg.HYPERP_OUTPUT_EMB,
29
  )
 
 
30
  anitag2vec.to(device)
31
+ anitag2vec.load_state_dict(torch.load(MODEL_PATH))
32
  anitag2vec.eval()
33
  runner = AniTag2VecRunner(tagtok, anitag2vec)
34
 
pytorch/anitag2vec_1a770feaa2bd9094_c7359727bcee4f8b_i128_e10_s60203_b300_p3841472.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:815e3f012aa2286131a015474c14607b8979654f73abe1eac7d5d4a69c71478d
3
+ size 15386061
pytorch/anitag2vec_1a770feaa2bd9094_c7359727bcee4f8b_i128_e15_s60203_b300_p3841472.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a456120b0678a8410c96e72a13abd3c85ed9a7f22b8a9575090af851ed9eb2e5
3
+ size 15386061
pytorch/anitag2vec_1a770feaa2bd9094_c7359727bcee4f8b_i128_e5_s60203_b300_p3841472.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cb4f4c402d8d8c42c793a3c5b6a9c0c6fd70c01f266899682c8f0f697755fe3f
3
+ size 15386018
pytorch/{anitag2vec_e15_s50000_p1871744.pth → anitag2vec_8ea07c7d34b64b69_c7359727bcee4f8b_i128_e10_s60203_b256_p1871744.pth} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2eb95a0d7a89a504bc38af484432792fa08f6d8199a7b703539156cab4a4cbe5
3
- size 7497806
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc0f3fc6c5c3bbac0b13832680bb361ab5c5ab1b2b260e627451488dcd26a680
3
+ size 7501026
pytorch/anitag2vec_8ea07c7d34b64b69_c7359727bcee4f8b_i128_e15_s60203_b256_p1871744.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1b74b574854e49022ba9416e0be18da3030f1b63141dbd6d645dff828810d20
3
+ size 7501026
pytorch/anitag2vec_8ea07c7d34b64b69_c7359727bcee4f8b_i128_e20_s60203_b256_p1871744.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0aac4b1858ca3c512c2d4175fa0fae7304935135a3c1cf330f3882b218a7f811
3
+ size 7501026
pytorch/anitag2vec_8ea07c7d34b64b69_c7359727bcee4f8b_i128_e5_s60203_b256_p1871744.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f1a3a89251213a2e295d3ca5620b2c6c89816b85c84d39a0820340d00ab7025
3
+ size 7500995
pytorch/anitag2vec_a75bfbdfe75fe264_c7359727bcee4f8b_i128_e10_s60203_b256_p1871744.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b3c3f887a2397f281b370ef3536f681597e3a2bd863c15c1383ee05563023d85
3
+ size 7501057
pytorch/anitag2vec_a75bfbdfe75fe264_c7359727bcee4f8b_i128_e15_s60203_b256_p1871744.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e7bd294eda9b3970ba940e1b95b21dd86ca17736625c42ee43cea103cb56ef4
3
+ size 7501057
pytorch/anitag2vec_a75bfbdfe75fe264_c7359727bcee4f8b_i128_e5_s60203_b256_p1871744.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba92f0c62d996509100d5c3905fe5deb1056f30852b3ee6c9f57f988897ed50c
3
+ size 7501026
pytorch/anitag2vec_e2ef15bc7697d18c_c7359727bcee4f8b_i128_e10_s60203_b100_p1871744.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:530247608e8529c387f8f2efe09b46d84f87c96611480214482d2d3c41fbd9c9
3
+ size 7501057
pytorch/anitag2vec_e2ef15bc7697d18c_c7359727bcee4f8b_i128_e15_s60203_b100_p1871744.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9edee4e6b36012d65d3ac3c12789fee5b15b6ce24fca3d30bfe0aef23aa29529
3
+ size 7501057
pytorch/anitag2vec_e2ef15bc7697d18c_c7359727bcee4f8b_i128_e5_s60203_b100_p1871744.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c8541749b6ed0ed88e029050f1559762fe398582db3f977a762a8cfada46092a
3
+ size 7501026
pytorch/errors_1a770feaa2bd9094_c7359727bcee4f8b.json ADDED
@@ -0,0 +1 @@
 
 
1
+ [1.5465073392758915, 1.191551031164862, 1.141250568539349, 1.114217244570528, 1.0815079971332455, 1.0874991256799271, 1.0734269378790215, 1.0518524273117977, 1.0536742379416282, 1.0285864217364373, 1.0393868693664892, 1.0260481107887345, 1.026582830580906, 1.0139693416173186, 1.012866616545625]
pytorch/errors_8ea07c7d34b64b69_c7359727bcee4f8b.json ADDED
@@ -0,0 +1 @@
 
 
1
+ [1.5338175347295857, 1.22657845560777, 1.143834860521858, 1.0836338509442442, 1.0620510950684547, 1.0459884273803841, 1.024043610666768, 1.0206486447635343, 1.0067960839655439, 1.016337098711628, 0.9958153492060758, 0.9780209674421003, 0.9733868704761489, 0.9744852811603223, 0.9751925069396779, 0.9733117154594195, 0.9643709912138471, 0.9583197345925589, 0.9579972770001929, 0.9580096569606813]
pytorch/errors_a75bfbdfe75fe264_c7359727bcee4f8b.json ADDED
@@ -0,0 +1 @@
 
 
1
+ [1.5629867716866024, 1.2090288374131009, 1.1101135576175432, 1.0536913489133626, 1.0466712425320834, 1.0251403781822173, 1.0098970648595842, 1.0102424606428309, 0.9888654357548488, 0.9933246557237738, 0.9835211908665754, 0.9751109461915695, 0.9704712323718152, 0.9701493646381265, 0.9687612371171935]
pytorch/errors_e2ef15bc7697d18c_c7359727bcee4f8b.json ADDED
@@ -0,0 +1 @@
 
 
1
+ [0.9241503570645825, 0.734504614527504, 0.6832865893438839, 0.6563414675744494, 0.6379766185772003, 0.6300161842127048, 0.6187568047279098, 0.6050779993306447, 0.6162416961177863, 0.5955178568771559, 0.5995774066960707, 0.5992496137232728, 0.5938771347100315, 0.5892666344180283, 0.5769423848360925]
pytorch/setup_params_1a770feaa2bd9094_c7359727bcee4f8b.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "TRAINING_TAKE_EXAMPLES": 70000,
3
+ "TRAINING_BATCH_SIZE": 300,
4
+ "TRAINING_PERM_LIMIT": 8,
5
+ "TRAINING_SUBARRAY_COUNT": 5,
6
+ "TRAINING_EPOCHS": 15,
7
+ "HYPERP_TAGTOK_MAX_TOKEN_CLAMP": 128,
8
+ "HYPERP_TAGTOK_VOCAB_SIZE": 5000,
9
+ "HYPERP_TAGTOK_MIN_FREQ": 3,
10
+ "HYPERP_TRANSFORMER_D_MODEL": 192,
11
+ "HYPERP_TRANSFORMER_N_HEADS": 6,
12
+ "HYPERP_TRANSFORMER_N_LAYERS": 3,
13
+ "HYPERP_OUTPUT_EMB": 128
14
+ }
pytorch/{setup_params.json → setup_params_8ea07c7d34b64b69_c7359727bcee4f8b.json} RENAMED
@@ -1,12 +1,14 @@
1
  {
2
- "TRAINING_TAKE_EXAMPLES": 50000,
3
  "TRAINING_BATCH_SIZE": 256,
 
 
 
4
  "HYPERP_TAGTOK_MAX_TOKEN_CLAMP": 128,
5
  "HYPERP_TAGTOK_VOCAB_SIZE": 5000,
6
  "HYPERP_TAGTOK_MIN_FREQ": 3,
7
  "HYPERP_TRANSFORMER_D_MODEL": 128,
8
  "HYPERP_TRANSFORMER_N_HEADS": 8,
9
  "HYPERP_TRANSFORMER_N_LAYERS": 2,
10
- "HYPERP_OUTPUT_EMB": 128,
11
- "HYPERP_EPOCHS": 15
12
  }
 
1
  {
2
+ "TRAINING_TAKE_EXAMPLES": 70000,
3
  "TRAINING_BATCH_SIZE": 256,
4
+ "TRAINING_PERM_LIMIT": 8,
5
+ "TRAINING_SUBARRAY_COUNT": 7,
6
+ "TRAINING_EPOCHS": 20,
7
  "HYPERP_TAGTOK_MAX_TOKEN_CLAMP": 128,
8
  "HYPERP_TAGTOK_VOCAB_SIZE": 5000,
9
  "HYPERP_TAGTOK_MIN_FREQ": 3,
10
  "HYPERP_TRANSFORMER_D_MODEL": 128,
11
  "HYPERP_TRANSFORMER_N_HEADS": 8,
12
  "HYPERP_TRANSFORMER_N_LAYERS": 2,
13
+ "HYPERP_OUTPUT_EMB": 128
 
14
  }
pytorch/setup_params_a75bfbdfe75fe264_c7359727bcee4f8b.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "TRAINING_TAKE_EXAMPLES": 70000,
3
+ "TRAINING_BATCH_SIZE": 256,
4
+ "TRAINING_PERM_LIMIT": 8,
5
+ "TRAINING_SUBARRAY_COUNT": 5,
6
+ "TRAINING_EPOCHS": 15,
7
+ "HYPERP_TAGTOK_MAX_TOKEN_CLAMP": 128,
8
+ "HYPERP_TAGTOK_VOCAB_SIZE": 5000,
9
+ "HYPERP_TAGTOK_MIN_FREQ": 3,
10
+ "HYPERP_TRANSFORMER_D_MODEL": 128,
11
+ "HYPERP_TRANSFORMER_N_HEADS": 8,
12
+ "HYPERP_TRANSFORMER_N_LAYERS": 2,
13
+ "HYPERP_OUTPUT_EMB": 128
14
+ }
pytorch/setup_params_e2ef15bc7697d18c_c7359727bcee4f8b.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "TRAINING_TAKE_EXAMPLES": 70000,
3
+ "TRAINING_BATCH_SIZE": 100,
4
+ "TRAINING_PERM_LIMIT": 8,
5
+ "TRAINING_SUBARRAY_COUNT": 5,
6
+ "TRAINING_EPOCHS": 15,
7
+ "HYPERP_TAGTOK_MAX_TOKEN_CLAMP": 128,
8
+ "HYPERP_TAGTOK_VOCAB_SIZE": 5000,
9
+ "HYPERP_TAGTOK_MIN_FREQ": 3,
10
+ "HYPERP_TRANSFORMER_D_MODEL": 128,
11
+ "HYPERP_TRANSFORMER_N_HEADS": 8,
12
+ "HYPERP_TRANSFORMER_N_LAYERS": 2,
13
+ "HYPERP_OUTPUT_EMB": 128
14
+ }
pytorch/{token_vocab_size_5000_freq_3.json → token_dataset_c7359727bcee4f8b_vocab_size_5000_freq_3.json} RENAMED
The diff for this file is too large to render. See raw diff