Upload 6 files

v1_inference/checkpoints/README.md

@@ -0,0 +1,9 @@
+# 权重放置说明
+
+权重不随仓库分发（微信等单独传递），放到本目录后按下表核对：
+
+| 文件 | 大小（字节） | md5 |
+|---|---|---|
+| fix_b3_1009_step74000.pt | 228027543 | b2571fb738eb7245bf7ddda4aa655308 |
+
+这是语音驱动 55D FLAME 的主模型，`runtime.py` 从 `checkpoints/fix_b3_1009_step74000.pt` 加载。

v1_inference/checkpoints/fix_b3_1009_step74000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1689980168ef574ce22f498ed6b6c3cbaf6b0e3863736bfaa28af8291d7a969c
+size 228027543

v1_inference/models/chinese-hubert-base/README.md

@@ -0,0 +1,11 @@
+# 权重放置说明
+
+HuBERT 特征提取器不随仓库分发（微信等单独传递），放到本目录后按下表核对：
+
+| 文件 | 大小（字节） | md5 |
+|---|---|---|
+| config.json | 1417 | 7663e254e02c3b634dc22e2f926ba368 |
+| preprocessor_config.json | 212 | baf88857eecfaf4bae54c9f85dbdaf49 |
+| pytorch_model.bin | 188811417 | 75d16b0997a5ba5ce75e8296fd83de65 |
+
+`runtime.py` 以 `local_files_only=True` 从本目录加载，三个文件缺一不可。本打包版只把 `config.json` 的 `_name_or_path` 清理为相对名称，不改变模型结构或权重。

v1_inference/models/chinese-hubert-base/config.json

@@ -0,0 +1,72 @@
+{
+  "_name_or_path": "chinese-hubert-base",
+  "activation_dropout": 0.1,
+  "apply_spec_augment": true,
+  "architectures": [
+    "HubertModel"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "conv_bias": false,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "sum",
+  "ctc_zero_infinity": false,
+  "do_stable_layer_norm": false,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_norm": "group",
+  "feat_proj_dropout": 0.0,
+  "feat_proj_layer_norm": true,
+  "final_dropout": 0.1,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.1,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
+  "model_type": "hubert",
+  "num_attention_heads": 12,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "torch_dtype": "float16",
+  "transformers_version": "4.30.2",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 32
+}

v1_inference/models/chinese-hubert-base/preprocessor_config.json

@@ -0,0 +1,9 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0,
+  "return_attention_mask": false,
+  "sampling_rate": 16000
+}

v1_inference/models/chinese-hubert-base/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24164f129c66499d1346e2aa55f183250c223161ec2770c0da3d3b08cf432d3c
+size 188811417