Hugging Face's logo

zuhri025
/
tts_weights_T2

TensorBoard
Safetensors
qwen3
Model card Files
xet
Metrics Community
zuhri025 commited on
Commit
8a6162e
·
verified ·
1 Parent(s): c5ba34e

Save checkpoint at step 13500

Browse files
Files changed (9) hide show
  1. checkpoint-13500/checkpoint_info.json +13 -0
  2. checkpoint-13500/config.json +52 -0
  3. checkpoint-13500/generation_config.json +9 -0
  4. checkpoint-13500/model.safetensors +3 -0
  5. checkpoint-13500/optimizer.pt +3 -0
  6. checkpoint-13500/tokenizer.json +0 -0
  7. checkpoint-13500/tokenizer_config.json +9 -0
  8. checkpoint-13500/training_config.json +55 -0
  9. checkpoint-13500/training_state.pt +3 -0
checkpoint-13500/checkpoint_info.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "step": 13500,
3
+ "best_val_loss": 10.131295871734618,
4
+ "tensorboard_logs": "tensorboard_logs",
5
+ "active_datasets": [
6
+ "globe",
7
+ "long_audio",
8
+ "emilia",
9
+ "gemini",
10
+ "soda"
11
+ ],
12
+ "total_samples": 672806
13
+ }
checkpoint-13500/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3ForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 3,
8
+ "dtype": "float32",
9
+ "eos_token_id": 3,
10
+ "head_dim": 128,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 512,
13
+ "initializer_range": 0.041666666666666664,
14
+ "intermediate_size": 2304,
15
+ "layer_types": [
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention"
33
+ ],
34
+ "max_position_embeddings": 1024,
35
+ "max_window_layers": 17,
36
+ "model_type": "qwen3",
37
+ "num_attention_heads": 4,
38
+ "num_hidden_layers": 17,
39
+ "num_key_value_heads": 1,
40
+ "pad_token_id": null,
41
+ "rms_norm_eps": 1e-06,
42
+ "rope_parameters": {
43
+ "rope_theta": 10000,
44
+ "rope_type": "default"
45
+ },
46
+ "sliding_window": null,
47
+ "tie_word_embeddings": false,
48
+ "transformers_version": "5.0.0",
49
+ "use_cache": true,
50
+ "use_sliding_window": false,
51
+ "vocab_size": 23947
52
+ }
checkpoint-13500/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 3,
4
+ "eos_token_id": 3,
5
+ "output_attentions": false,
6
+ "output_hidden_states": false,
7
+ "transformers_version": "5.0.0",
8
+ "use_cache": true
9
+ }
checkpoint-13500/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:db16126a1c7ca208b667a8d36451fd578b1d445e0afbd7927aac5a77e11b9359
3
+ size 383409712
checkpoint-13500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:350bfcf7a5d3b84060ad3efeb8c474f70e56b4ce740530d9852115688903ca49
3
+ size 766939851
checkpoint-13500/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-13500/tokenizer_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "bos_token": "<|bos|>",
4
+ "eos_token": "<|eos|>",
5
+ "model_max_length": 1000000000000000019884624838656,
6
+ "pad_token": "<|pad|>",
7
+ "tokenizer_class": "TokenizersBackend",
8
+ "unk_token": "<|unk|>"
9
+ }
checkpoint-13500/training_config.json ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "tokenizer_id": "zuhri025/TTS-Tokenizer-T",
3
+ "model_config_id": "ekwek/Soprano-1.1-80M",
4
+ "save_dir": "tts_weights_T2",
5
+ "repo_name": "tts_weights_T2",
6
+ "active_datasets": [
7
+ "globe",
8
+ "long_audio",
9
+ "emilia",
10
+ "gemini",
11
+ "soda"
12
+ ],
13
+ "dataset_limits": {
14
+ "gemini": 10000,
15
+ "globe": 250000,
16
+ "soda": 250000,
17
+ "emilia": 80000,
18
+ "long_audio": 70000
19
+ },
20
+ "max_samples": 100000,
21
+ "device": "cuda",
22
+ "num_workers": 2,
23
+ "pin_memory": true,
24
+ "compile_model": true,
25
+ "use_mixed_precision": true,
26
+ "seed": 1337,
27
+ "max_steps": 20000,
28
+ "batch_size": 2,
29
+ "grad_accum_steps": 4,
30
+ "seq_len": 1024,
31
+ "max_lr": 0.001,
32
+ "min_lr": 5e-05,
33
+ "warmup_ratio": 0.1,
34
+ "cooldown_ratio": 0.1,
35
+ "betas": [
36
+ 0.9,
37
+ 0.95
38
+ ],
39
+ "weight_decay": 0.1,
40
+ "grad_clip_norm": 1.0,
41
+ "text_loss_weight": 0.5,
42
+ "text_mode_ratio": 1.0,
43
+ "phoneme_mode_ratio": 0.0,
44
+ "mix_mode_ratio": 0.0,
45
+ "unconditional_prob": 0.05,
46
+ "voice_dropout_prob": 0.1,
47
+ "word_mix_ratio": 0.5,
48
+ "val_freq": 1500,
49
+ "save_freq": 1500,
50
+ "log_freq": 10,
51
+ "val_steps": 20,
52
+ "test_split_size": 0.01,
53
+ "upload_tensorboard": true,
54
+ "tensorboard_dir": "tensorboard_logs"
55
+ }
checkpoint-13500/training_state.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8df70d3c97be3e31bd6480f04dd24cb5b7dc1e34c10eea1dbca1f35ac45070c4
3
+ size 766942923