Upload UltravoxPipeline
Browse files- config.json +4 -10
- generation_config.json +2 -1
- special_tokens_map.json +7 -1
- ultravox_pipeline.py +5 -1
config.json
CHANGED
@@ -154,16 +154,10 @@
|
|
154 |
},
|
155 |
"custom_pipelines": {
|
156 |
"ultravox-pipeline": {
|
157 |
-
"default": {
|
158 |
-
"model": {
|
159 |
-
"pt": [
|
160 |
-
"fixie-ai/ultravox-v0.2",
|
161 |
-
"main"
|
162 |
-
]
|
163 |
-
}
|
164 |
-
},
|
165 |
"impl": "ultravox_pipeline.UltravoxPipeline",
|
166 |
-
"pt": [
|
|
|
|
|
167 |
"tf": [],
|
168 |
"type": "multimodal"
|
169 |
}
|
@@ -193,6 +187,6 @@
|
|
193 |
},
|
194 |
"text_model_id": null,
|
195 |
"torch_dtype": "bfloat16",
|
196 |
-
"transformers_version": "4.41.
|
197 |
"vocab_size": 128256
|
198 |
}
|
|
|
154 |
},
|
155 |
"custom_pipelines": {
|
156 |
"ultravox-pipeline": {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
157 |
"impl": "ultravox_pipeline.UltravoxPipeline",
|
158 |
+
"pt": [
|
159 |
+
"AutoModel"
|
160 |
+
],
|
161 |
"tf": [],
|
162 |
"type": "multimodal"
|
163 |
}
|
|
|
187 |
},
|
188 |
"text_model_id": null,
|
189 |
"torch_dtype": "bfloat16",
|
190 |
+
"transformers_version": "4.41.1",
|
191 |
"vocab_size": 128256
|
192 |
}
|
generation_config.json
CHANGED
@@ -2,5 +2,6 @@
|
|
2 |
"_from_model_config": true,
|
3 |
"bos_token_id": 128000,
|
4 |
"eos_token_id": 128009,
|
5 |
-
"
|
|
|
6 |
}
|
|
|
2 |
"_from_model_config": true,
|
3 |
"bos_token_id": 128000,
|
4 |
"eos_token_id": 128009,
|
5 |
+
"pad_token_id": 128009,
|
6 |
+
"transformers_version": "4.41.1"
|
7 |
}
|
special_tokens_map.json
CHANGED
@@ -13,5 +13,11 @@
|
|
13 |
"rstrip": false,
|
14 |
"single_word": false
|
15 |
},
|
16 |
-
"pad_token":
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
}
|
|
|
13 |
"rstrip": false,
|
14 |
"single_word": false
|
15 |
},
|
16 |
+
"pad_token": {
|
17 |
+
"content": "<|eot_id|>",
|
18 |
+
"lstrip": false,
|
19 |
+
"normalized": false,
|
20 |
+
"rstrip": false,
|
21 |
+
"single_word": false
|
22 |
+
}
|
23 |
}
|
ultravox_pipeline.py
CHANGED
@@ -67,11 +67,15 @@ class UltravoxPipeline(transformers.Pipeline):
|
|
67 |
"No sampling rate provided, using default of 16kHz. We highly recommend providing the correct sampling rate."
|
68 |
)
|
69 |
|
70 |
-
|
71 |
text=text,
|
72 |
audio=inputs["audio"],
|
73 |
sampling_rate=inputs.get("sampling_rate", 16000),
|
74 |
)
|
|
|
|
|
|
|
|
|
75 |
|
76 |
def _forward(
|
77 |
self,
|
|
|
67 |
"No sampling rate provided, using default of 16kHz. We highly recommend providing the correct sampling rate."
|
68 |
)
|
69 |
|
70 |
+
output = self.processor(
|
71 |
text=text,
|
72 |
audio=inputs["audio"],
|
73 |
sampling_rate=inputs.get("sampling_rate", 16000),
|
74 |
)
|
75 |
+
if "audio_values" in output:
|
76 |
+
output["audio_values"] = output["audio_values"].to(self.model.dtype)
|
77 |
+
|
78 |
+
return output
|
79 |
|
80 |
def _forward(
|
81 |
self,
|