Update tiny models for FunAsrNanoModel

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

chat_template.jinja

@@ -0,0 +1,15 @@
+{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system
+You are a helpful assistant.<|im_end|>
+{% endif %}<|im_start|>{{ message['role'] }}
+{% if message['content'] is string %}{{ message['content'] }}<|im_end|>
+{% else %}{% set audio_items = message['content'] | selectattr('type', 'equalto', 'audio') | list %}{% if message['role'] == 'user' and audio_items %}{% set text_items = message['content'] | selectattr('type', 'equalto', 'text') | list %}{% set keyword_items = message['content'] | selectattr('type', 'equalto', 'keywords') | list %}{% set language_items = message['content'] | selectattr('type', 'equalto', 'language') | list %}{% if text_items or keyword_items %}请结合上下文信息，更加准确地完成语音转写任务。如果没有相关信息，我们会留空。
+
+
+**上下文信息：**
+
+
+{% for item in text_items %}{{ item['text'] }}
+{% endfor %}{% set keyword_namespace = namespace(items=[]) %}{% for item in keyword_items %}{% set keyword_namespace.items = keyword_namespace.items + item['keywords'] %}{% endfor %}{% if keyword_namespace.items %}热词列表：[{{ keyword_namespace.items | join(', ') }}]
+{% endif %}{% endif %}语音转写{% if language_items %}成{{ language_items[0]['language'] }}{% endif %}：{% for content in audio_items %}<|object_ref_start|>{% endfor %}{% else %}{% for content in message['content'] %}{% if content['type'] == 'audio' %}<|object_ref_start|>{% elif content['type'] == 'text' %}{{ content['text'] }}{% endif %}{% endfor %}{% endif %}<|im_end|>
+{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant
+{% endif %}

config.json

@@ -0,0 +1,73 @@
+{
+  "adaptor_config": {
+    "attention_dropout": 0.0,
+    "hidden_act": "relu",
+    "hidden_dropout": 0.0,
+    "hidden_size": 64,
+    "intermediate_size": 16,
+    "layer_norm_eps": 1e-05,
+    "model_type": "",
+    "num_attention_heads": 4,
+    "num_hidden_layers": 2,
+    "projector_hidden_act": "relu",
+    "projector_hidden_size": 64
+  },
+  "architectures": [
+    "FunAsrNanoModel"
+  ],
+  "audio_config": {
+    "attention_dropout": 0.1,
+    "fsmn_kernel_size": 5,
+    "hidden_act": "gelu",
+    "hidden_dropout": 0.0,
+    "hidden_size": 64,
+    "intermediate_size": 64,
+    "layer_norm_eps": 1e-05,
+    "max_position_embeddings": 128,
+    "model_type": "fun_asr_nano_encoder",
+    "num_attention_heads": 4,
+    "num_hidden_layers": 2,
+    "num_mel_bins": 80,
+    "num_stacked_frames": 1,
+    "num_timestamp_prediction_layers": 1
+  },
+  "audio_token_id": 0,
+  "dtype": "float32",
+  "initializer_range": 0.02,
+  "model_type": "fun_asr_nano",
+  "pad_token_id": 151643,
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 1,
+    "eos_token_id": 151645,
+    "head_dim": 16,
+    "hidden_act": "gelu",
+    "hidden_size": 64,
+    "initializer_range": 0.02,
+    "intermediate_size": 64,
+    "layer_types": [
+      "full_attention",
+      "full_attention"
+    ],
+    "max_position_embeddings": 200,
+    "max_window_layers": 28,
+    "model_type": "qwen3",
+    "num_attention_heads": 4,
+    "num_hidden_layers": 2,
+    "num_key_value_heads": 2,
+    "pad_token_id": 151643,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    },
+    "sliding_window": null,
+    "tie_word_embeddings": false,
+    "use_cache": true,
+    "use_sliding_window": false,
+    "vocab_size": 151669
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.19.0.dev0"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9eee82789b7b2b3ef1a5919bfba94f70b57c13b227711957bfa372859c380caa
+size 39439616

preprocessor_config.json

@@ -0,0 +1,12 @@
+{
+  "feature_extractor_type": "FunAsrNanoFeatureExtractor",
+  "feature_size": 80,
+  "frame_length": 25,
+  "frame_shift": 10,
+  "num_frames_lfr": 7,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": true,
+  "sampling_rate": 16000,
+  "stride_lfr": 6
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
+size 11422650

tokenizer_config.json

@@ -0,0 +1,33 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": true,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding": true,
+  "padding_side": "left",
+  "processor_class": "FunAsrNanoProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}