Hugging Face's logo

saryps-labs
/
indicF5-rt

Text-to-Speech
ONNX
Model card Files
xet
Community
1
Sedherthe commited on
Commit
0721ed8
·
verified ·
1 Parent(s): ec4142b

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +175 -0
config.json ADDED
@@ -0,0 +1,175 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "0.15.0",
3
+ "pretrained_config": {
4
+ "architecture": "F5TTS",
5
+ "dtype": "float16",
6
+ "vocab_size": null,
7
+ "hidden_size": 1024,
8
+ "num_hidden_layers": 22,
9
+ "num_attention_heads": 16,
10
+ "hidden_act": "gelu",
11
+ "logits_dtype": "float32",
12
+ "norm_epsilon": 1e-05,
13
+ "runtime_defaults": null,
14
+ "position_embedding_type": "learned_absolute",
15
+ "max_position_embeddings": null,
16
+ "num_key_value_heads": 16,
17
+ "intermediate_size": 4096,
18
+ "mapping": {
19
+ "world_size": 1,
20
+ "gpus_per_node": 8,
21
+ "cp_size": 1,
22
+ "tp_size": 1,
23
+ "pp_size": 1,
24
+ "moe_tp_size": 1,
25
+ "moe_ep_size": 1
26
+ },
27
+ "quantization": {
28
+ "quant_algo": null,
29
+ "kv_cache_quant_algo": null,
30
+ "group_size": 128,
31
+ "smoothquant_val": 0.5,
32
+ "clamp_val": null,
33
+ "use_meta_recipe": false,
34
+ "has_zero_point": false,
35
+ "pre_quant_scale": false,
36
+ "exclude_modules": null
37
+ },
38
+ "use_parallel_embedding": false,
39
+ "embedding_sharding_dim": 0,
40
+ "share_embedding_table": false,
41
+ "head_size": 64,
42
+ "qk_layernorm": false,
43
+ "dim_head": 64,
44
+ "dropout": 0.1,
45
+ "ff_mult": 2,
46
+ "mel_dim": 100,
47
+ "text_num_embeds": 256,
48
+ "text_dim": 512,
49
+ "conv_layers": 4,
50
+ "long_skip_connection": false
51
+ },
52
+ "build_config": {
53
+ "max_input_len": 1024,
54
+ "max_seq_len": 2048,
55
+ "opt_batch_size": 8,
56
+ "max_batch_size": 2048,
57
+ "max_beam_width": 1,
58
+ "max_num_tokens": 4194304,
59
+ "opt_num_tokens": null,
60
+ "max_prompt_embedding_table_size": 0,
61
+ "kv_cache_type": "PAGED",
62
+ "gather_context_logits": false,
63
+ "gather_generation_logits": false,
64
+ "strongly_typed": true,
65
+ "force_num_profiles": null,
66
+ "profiling_verbosity": "layer_names_only",
67
+ "enable_debug_output": false,
68
+ "max_draft_len": 0,
69
+ "speculative_decoding_mode": 1,
70
+ "use_refit": false,
71
+ "input_timing_cache": null,
72
+ "output_timing_cache": "model.cache",
73
+ "lora_config": {
74
+ "lora_dir": [],
75
+ "lora_ckpt_source": "hf",
76
+ "max_lora_rank": 64,
77
+ "lora_target_modules": [],
78
+ "trtllm_modules_to_hf_modules": {}
79
+ },
80
+ "auto_parallel_config": {
81
+ "world_size": 1,
82
+ "gpus_per_node": 8,
83
+ "cluster_key": "Tesla-T4",
84
+ "cluster_info": {
85
+ "inter_node_bw_per_device": 25,
86
+ "intra_node_bw_per_device": 5,
87
+ "inter_node_latency": 10,
88
+ "intra_node_latency": 10,
89
+ "intra_node_sharp": false,
90
+ "inter_node_sharp": true,
91
+ "memory_bw": 320,
92
+ "memory_budget_per_device": 15,
93
+ "math_throughput": {
94
+ "int4": 260,
95
+ "int8": 130,
96
+ "fp8": 0,
97
+ "float16": 65,
98
+ "bfloat16": 0,
99
+ "float32": 8
100
+ },
101
+ "memory_efficiency": 1.0,
102
+ "math_efficiency": 1.0,
103
+ "communication_efficiency": 1.0
104
+ },
105
+ "sharding_cost_model": "alpha_beta",
106
+ "comm_cost_model": "alpha_beta",
107
+ "enable_pipeline_parallelism": false,
108
+ "enable_shard_unbalanced_shape": false,
109
+ "enable_shard_dynamic_shape": false,
110
+ "enable_reduce_scatter": true,
111
+ "builder_flags": null,
112
+ "debug_mode": false,
113
+ "infer_shape": true,
114
+ "validation_mode": false,
115
+ "same_buffer_io": {
116
+ "past_key_value_(\\d+)": "present_key_value_\\1"
117
+ },
118
+ "same_spec_io": {},
119
+ "sharded_io_allowlist": [
120
+ "past_key_value_\\d+",
121
+ "present_key_value_\\d*"
122
+ ],
123
+ "fill_weights": false,
124
+ "parallel_config_cache": null,
125
+ "profile_cache": null,
126
+ "dump_path": null,
127
+ "debug_outputs": []
128
+ },
129
+ "weight_sparsity": false,
130
+ "weight_streaming": false,
131
+ "plugin_config": {
132
+ "dtype": "float16",
133
+ "bert_attention_plugin": null,
134
+ "gpt_attention_plugin": "auto",
135
+ "gemm_plugin": null,
136
+ "gemm_swiglu_plugin": null,
137
+ "fp8_rowwise_gemm_plugin": null,
138
+ "smooth_quant_gemm_plugin": null,
139
+ "qserve_gemm_plugin": null,
140
+ "identity_plugin": null,
141
+ "layernorm_quantization_plugin": null,
142
+ "rmsnorm_quantization_plugin": null,
143
+ "nccl_plugin": null,
144
+ "lora_plugin": null,
145
+ "weight_only_groupwise_quant_matmul_plugin": null,
146
+ "weight_only_quant_matmul_plugin": null,
147
+ "smooth_quant_plugins": true,
148
+ "quantize_per_token_plugin": false,
149
+ "quantize_tensor_plugin": false,
150
+ "moe_plugin": "auto",
151
+ "mamba_conv1d_plugin": "auto",
152
+ "low_latency_gemm_plugin": null,
153
+ "low_latency_gemm_swiglu_plugin": null,
154
+ "context_fmha": true,
155
+ "bert_context_fmha_fp32_acc": false,
156
+ "paged_kv_cache": true,
157
+ "remove_input_padding": false,
158
+ "reduce_fusion": false,
159
+ "enable_xqa": true,
160
+ "tokens_per_block": 64,
161
+ "use_paged_context_fmha": false,
162
+ "use_fp8_context_fmha": false,
163
+ "multiple_profiles": false,
164
+ "paged_state": false,
165
+ "streamingllm": false,
166
+ "manage_weights": false,
167
+ "use_fused_mlp": true,
168
+ "pp_reduce_scatter": false
169
+ },
170
+ "use_strip_plan": false,
171
+ "max_encoder_input_len": 1024,
172
+ "use_fused_mlp": true,
173
+ "monitor_memory": false
174
+ }
175
+ }