Upload folder using huggingface_hub

.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

README.md

@@ -0,0 +1,136 @@
+---
+library_name: transformers
+base_model:
+- k2-fsa/OmniVoice
+---
+
+This tiny model is intended for debugging. It is randomly initialized using the configuration adapted from [k2-fsa/OmniVoice](https://huggingface.co/k2-fsa/OmniVoice).
+
+| File path | Size |
+|------|------|
+| model.safetensors | 5.5MB |
+| audio_tokenizer/model.safetensors | 6.7MB |
+
+
+### Example usage:
+
+```python
+from omnivoice import OmniVoice
+import torch
+import torchaudio
+
+model_id = "tiny-random/omnivoice"
+model = OmniVoice.from_pretrained(
+    model_id,
+    dtype=torch.bfloat16,
+)
+audio = model.generate(
+    text="Hello, this is test example 1",
+    instruct="low pitch, british accent",
+)
+torchaudio.save("/tmp/example1.wav", audio[0], 24000)
+
+audio2 = model.generate(
+    text="Hello, this is test example 2",
+    ref_audio="/tmp/example1.wav",
+    ref_text="Hello, this is test example 1",
+)
+torchaudio.save("/tmp/example2.wav", audio2[0], 24000)
+```
+
+### Codes to create this repo:
+
+<details>
+<summary>Click to expand</summary>
+
+```python
+import torch
+import os
+
+from transformers import (
+    set_seed,
+    AutoConfig,
+    AutoTokenizer,
+    HiggsAudioV2TokenizerModel,
+    AutoFeatureExtractor,
+)
+from huggingface_hub import hf_hub_download
+import json
+from omnivoice import OmniVoice, OmniVoiceConfig
+
+source_model_id = "k2-fsa/OmniVoice"
+save_folder = "/tmp/tiny-random/omnivoice"
+
+set_seed(42)
+tokenizer = AutoTokenizer.from_pretrained(source_model_id, trust_remote_code=True)
+tokenizer.save_pretrained(save_folder)
+
+with open(
+    hf_hub_download(source_model_id, filename="audio_tokenizer/config.json", repo_type="model"),
+    "r",
+    encoding="utf-8",
+) as f:
+    config_dict = json.load(f)
+config_dict["acoustic_model_config"].update(
+    {
+        "decoder_hidden_size": 32,
+        "encoder_hidden_size": 4,
+        "hidden_size": 4,
+        "codebook_dim": 8,
+    }
+)
+config_dict["semantic_model_config"].update(
+    {
+        "conv_dim": [8] * 7,
+        "hidden_size": 16 * 4,
+        "intermediate_size": 64,
+        "num_attention_heads": 4,
+        "num_hidden_layers": 2,
+    }
+)
+os.makedirs(os.path.join(save_folder, "audio_tokenizer"), exist_ok=True)
+with open(os.path.join(save_folder, "audio_tokenizer/config.json"), "w", encoding="utf-8") as f:
+    json.dump(config_dict, f, ensure_ascii=False, indent=2)
+audio_tokenizer = HiggsAudioV2TokenizerModel(
+    AutoConfig.from_pretrained(os.path.join(save_folder, "audio_tokenizer"))
+)
+audio_tokenizer.save_pretrained(os.path.join(save_folder, "audio_tokenizer"))
+print(audio_tokenizer)
+set_seed(42)
+with torch.no_grad():
+    for name, p in sorted(audio_tokenizer.named_parameters()):
+        torch.nn.init.normal_(p, 0, 0.2)
+        print(name, p.shape)
+
+feature_extractor = AutoFeatureExtractor.from_pretrained(source_model_id, subfolder="audio_tokenizer")
+feature_extractor.save_pretrained(os.path.join(save_folder, "audio_tokenizer"))
+
+with open(
+    hf_hub_download(source_model_id, filename="config.json", repo_type="model"),
+    "r",
+    encoding="utf-8",
+) as f:
+    config_dict = json.load(f)
+config_dict["llm_config"].update(
+    {
+        "hidden_size": 8,
+        "head_dim": 32,
+        "intermediate_size": 32,
+        "num_attention_heads": 8,
+        "num_key_value_heads": 4,
+        "num_hidden_layers": 4,
+        "max_window_layers": 2,
+        "layer_types": ["full_attention"] * 4,
+    }
+)
+config = OmniVoiceConfig.from_dict(config_dict)
+model = OmniVoice(config).eval()
+set_seed(42)
+with torch.no_grad():
+    for name, p in sorted(model.named_parameters()):
+        torch.nn.init.normal_(p, 0, 0.2)
+        print(name, p.shape)
+model.save_pretrained(save_folder)
+```
+
+</details>

audio_tokenizer/config.json

@@ -0,0 +1,129 @@
+{
+  "acoustic_model_config": {
+    "codebook_dim": 8,
+    "codebook_loss_weight": 1.0,
+    "codebook_size": 1024,
+    "commitment_loss_weight": 0.25,
+    "decoder_hidden_size": 32,
+    "downsampling_ratios": [
+      8,
+      5,
+      4,
+      2,
+      3
+    ],
+    "encoder_hidden_size": 4,
+    "hidden_size": 4,
+    "hop_length": 960,
+    "model_type": "dac",
+    "n_codebooks": 9,
+    "quantizer_dropout": 0,
+    "sampling_rate": 16000,
+    "upsampling_ratios": [
+      8,
+      5,
+      4,
+      2,
+      3
+    ]
+  },
+  "architectures": [
+    "HiggsAudioV2TokenizerModel"
+  ],
+  "block_dilations": [
+    1,
+    1
+  ],
+  "channel_ratios": [
+    1,
+    1
+  ],
+  "codebook_dim": 64,
+  "codebook_size": 1024,
+  "downsample_factor": 320,
+  "dtype": "float32",
+  "initializer_range": 0.02,
+  "kernel_size": 3,
+  "model_type": "higgs_audio_v2_tokenizer",
+  "sample_rate": 24000,
+  "semantic_model_config": {
+    "activation_dropout": 0.1,
+    "apply_spec_augment": true,
+    "attention_dropout": 0.1,
+    "bos_token_id": 1,
+    "classifier_proj_size": 256,
+    "conv_bias": false,
+    "conv_dim": [
+      8,
+      8,
+      8,
+      8,
+      8,
+      8,
+      8
+    ],
+    "conv_kernel": [
+      10,
+      3,
+      3,
+      3,
+      3,
+      2,
+      2
+    ],
+    "conv_pos_batch_norm": false,
+    "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": 64,
+    "initializer_range": 0.02,
+    "intermediate_size": 64,
+    "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": 0.0,
+    "model_type": "hubert",
+    "num_attention_heads": 4,
+    "num_conv_pos_embedding_groups": 16,
+    "num_conv_pos_embeddings": 128,
+    "num_feat_extract_layers": 7,
+    "num_hidden_layers": 2,
+    "pad_token_id": 0,
+    "use_weighted_layer_sum": false,
+    "vocab_size": 32
+  },
+  "semantic_sample_rate": 16000,
+  "strides": [
+    1,
+    1
+  ],
+  "target_bandwidths": [
+    0.5,
+    1,
+    1.5,
+    2
+  ],
+  "transformers_version": "5.5.0",
+  "unit_kernel_size": 3
+}

audio_tokenizer/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0478f3bc248298916cb9fc1f9acfae4d39dc325ca337830340911179969ce50
+size 6666464

audio_tokenizer/preprocessor_config.json

@@ -0,0 +1,9 @@
+{
+  "feature_extractor_type": "DacFeatureExtractor",
+  "feature_size": 1,
+  "hop_length": 960,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": true,
+  "sampling_rate": 24000
+}

chat_template.jinja

@@ -0,0 +1,89 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

config.json

@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "OmniVoice"
+  ],
+  "audio_codebook_weights": [
+    8,
+    8,
+    6,
+    6,
+    4,
+    4,
+    2,
+    2
+  ],
+  "audio_mask_id": 1024,
+  "audio_vocab_size": 1025,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": 151645,
+  "llm_config": {
+    "_name_or_path": "",
+    "architectures": [
+      "Qwen3ForCausalLM"
+    ],
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 151643,
+    "chunk_size_feed_forward": 0,
+    "dtype": "float32",
+    "eos_token_id": 151645,
+    "head_dim": 32,
+    "hidden_act": "silu",
+    "hidden_size": 8,
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "initializer_range": 0.02,
+    "intermediate_size": 32,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "layer_types": [
+      "full_attention",
+      "full_attention",
+      "full_attention",
+      "full_attention"
+    ],
+    "max_position_embeddings": 40960,
+    "max_window_layers": 2,
+    "model_type": "qwen3",
+    "num_attention_heads": 8,
+    "num_hidden_layers": 4,
+    "num_key_value_heads": 4,
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "pad_token_id": null,
+    "problem_type": null,
+    "return_dict": true,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "rope_theta": 1000000,
+      "rope_type": "default"
+    },
+    "sliding_window": null,
+    "tie_word_embeddings": true,
+    "use_cache": true,
+    "use_sliding_window": false,
+    "vocab_size": 151676
+  },
+  "model_type": "omnivoice",
+  "num_audio_codebook": 8,
+  "pad_token_id": 151643,
+  "transformers_version": "5.5.0"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d626de7b879f2d8cf439f1627481afbfd67ad288a67d59cc54dce41bcea5fc62
+size 5495352

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:408f669b7e2b045fdf54201d815bd364e6667dbd845115da81239c40bc6dcfd1
+size 11423986

tokenizer_config.json

@@ -0,0 +1,23 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|denoise|>",
+    "<|lang_start|>",
+    "<|lang_end|>",
+    "<|instruct_start|>",
+    "<|instruct_end|>",
+    "<|text_start|>",
+    "<|text_end|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}