Upload folder using huggingface_hub

config.json

@@ -42,16 +42,16 @@
   "text_config": {
     "attention_dropout": 0.0,
     "hidden_act": "silu",
-    "hidden_size": 2048,
+    "hidden_size": 3584,
     "init_std": 0.02,
     "initializer_range": 0.02,
-    "intermediate_size": 11008,
+    "intermediate_size": 18944,
     "max_position_embeddings": 32768,
-    "max_window_layers": 70,
+    "max_window_layers": 28,
     "model_type": "qwen2_5_omni_text",
-    "num_attention_heads": 16,
-    "num_hidden_layers": 36,
-    "num_key_value_heads": 2,
+    "num_attention_heads": 28,
+    "num_hidden_layers": 28,
+    "num_key_value_heads": 4,
     "rms_norm_eps": 1e-06,
     "rope_scaling": {
       "mrope_section": [
@@ -66,7 +66,7 @@
     "sliding_window": 32768,
     "use_cache": true,
     "use_sliding_window": false,
-    "vocab_size": 151936
+    "vocab_size": 152064
   },
   "torch_dtype": "float32",
   "transformers_version": "4.52.4"

model-00001-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:324d47a69b306b736f9c1ed9c3ac6b2f08dd25f3238e0995ca03d1f628d14d3f
+size 4962055488

model-00002-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc9143a966742e2d38c7039d905d907211ec9b1d504786dc6cfa1655db281405
+size 4932744424

model-00003-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61fbad4004a9b2bf23212617ddbdc92b37776a208feb85233cadac3b48cda9cf
+size 4932744424

model-00004-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:363674f4ba1fbd50dfe599ddde8c687cbb4ae95f49b56bb41ca842a60dc8fc06
+size 4998852848

model-00005-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf4af607c8914b24b50f4adbc0e919bfb5ad8cf909d0ba02aa1145e2ee3c706e
+size 4984124832

model-00006-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:439ab29e80c237c362d9aa58f03e1fa6311ca0513613f26c077b15623bdeca22
+size 4932744480

model-00007-of-00007.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d80f1068c04d6a50bc627b0e97a30ea3b97ce0609284a174a4f9795e77bf1818
+size 3383846936

modeling_midashenglm.py

@@ -1,18 +1,7 @@
 import collections
 import collections.abc
 from dataclasses import dataclass
-from typing import (
-    Any,
-    Callable,
-    Iterable,
-    List,
-    Optional,
-    Sequence,
-    Tuple,
-    Union,
-    Unpack,
-    cast,
-)
+from typing import Any, Callable, Iterable, List, Optional, Sequence, Tuple, Union, cast
 
 import torch
 import torch.nn as nn
@@ -28,6 +17,7 @@ from transformers.models.qwen2_5_omni.modeling_qwen2_5_omni import (
     Qwen2_5OmniThinkerTextModel,
 )
 from transformers.utils import LossKwargs, can_return_tuple
+from typing_extensions import Unpack
 
 from .configuration_midashenglm import DashengConfig, MiDashengLMConfig
 

processing_midashenglm.py

@@ -4,7 +4,8 @@ import numpy as np
 import torch
 from transformers import Qwen2Tokenizer, Qwen2TokenizerFast, Wav2Vec2FeatureExtractor
 from transformers.feature_extraction_utils import BatchFeature
-from transformers.processing_utils import ProcessingKwargs, ProcessorMixin, Unpack
+from transformers.processing_utils import ProcessingKwargs, ProcessorMixin
+from typing_extensions import Unpack
 
 
 class MiDashengLMProcessorKwargs(ProcessingKwargs):