facebook

/

s2t-small-covost2-fr-en-st

like 0

Follow

AI at Meta 4.28k

When I run

import torch
from transformers import Speech2TextProcessor, Speech2TextForConditionalGeneration
from datasets import load_dataset
import soundfile as sf

model = Speech2TextForConditionalGeneration.from_pretrained("facebook/s2t-small-covost2-fr-en-st")
processor = Speech2TextProcessor.from_pretrained("facebook/s2t-small-covost2-fr-en-st")

def map_to_array(batch):
    speech, _ = sf.read(batch["file"])
    batch["speech"] = speech
    return batch

ds = load_dataset(
    "patrickvonplaten/librispeech_asr_dummy",
    "clean",
    split="validation"
)
ds = ds.map(map_to_array)

inputs = processor(
    ds["speech"][0],
    sampling_rate=48_000,
    return_tensors="pt"
)
generated_ids = model.generate(input_ids=inputs["input_features"], attention_mask=inputs["attention_mask"])

translation = processor.batch_decode(generated_ids, skip_special_tokens=True)

I get the following error.

(env) quentingallouedec@MacBook-Pro-de-Quentin gia % /Users/quentingallouedec/gia/env/bin/python /Users/qu
entingallouedec/gia/examples/s2t.py
WARNING:datasets.builder:Found cached dataset librispeech_asr_dummy (/Users/quentingallouedec/.cache/huggingface/datasets/patrickvonplaten___librispeech_asr_dummy/clean/2.1.0/f2c70a4d03ab4410954901bde48c54b85ca1b7f9bf7d616e7e2a72b5ee6ddbfc)
WARNING:datasets.arrow_dataset:Loading cached processed dataset at /Users/quentingallouedec/.cache/huggingface/datasets/patrickvonplaten___librispeech_asr_dummy/clean/2.1.0/f2c70a4d03ab4410954901bde48c54b85ca1b7f9bf7d616e7e2a72b5ee6ddbfc/cache-2944a8173f386e6d.arrow
╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│                                                                                                  │
│ /Users/quentingallouedec/gia/examples/s2t.py:26 in <module>                                      │
│                                                                                                  │
│   23 │   sampling_rate=48_000,                                                                   │
│   24 │   return_tensors="pt"                                                                     │
│   25 )                                                                                           │
│ ❱ 26 generated_ids = model.generate(input_ids=inputs["input_features"], attention_mask=inputs    │
│   27                                                                                             │
│   28 translation = processor.batch_decode(generated_ids, skip_special_tokens=True)               │
│   29                                                                                             │
│ /Users/quentingallouedec/gia/env/lib/python3.10/site-packages/torch/utils/_contextlib.py:115 in  │
│ decorate_context                                                                                 │
│                                                                                                  │
│   112 │   @functools.wraps(func)                                                                 │
│   113 │   def decorate_context(*args, **kwargs):                                                 │
│   114 │   │   with ctx_factory():                                                                │
│ ❱ 115 │   │   │   return func(*args, **kwargs)                                                   │
│   116 │                                                                                          │
│   117 │   return decorate_context                                                                │
│   118                                                                                            │
│                                                                                                  │
│ /Users/quentingallouedec/gia/env/lib/python3.10/site-packages/transformers/generation/utils.py:1 │
│ 197 in generate                                                                                  │
│                                                                                                  │
│   1194 │   │                                                                                     │
│   1195 │   │   generation_config = copy.deepcopy(generation_config)                              │
│   1196 │   │   model_kwargs = generation_config.update(**kwargs)  # All unused kwargs must be m  │
│ ❱ 1197 │   │   self._validate_model_kwargs(model_kwargs.copy())                                  │
│   1198 │   │                                                                                     │
│   1199 │   │   # 2. Set generation parameters if not already defined                             │
│   1200 │   │   logits_processor = logits_processor if logits_processor is not None else LogitsP  │
│                                                                                                  │
│ /Users/quentingallouedec/gia/env/lib/python3.10/site-packages/transformers/generation/utils.py:1 │
│ 090 in _validate_model_kwargs                                                                    │
│                                                                                                  │
│   1087 │   │   │   │   unused_model_args.append(key)                                             │
│   1088 │   │                                                                                     │
│   1089 │   │   if unused_model_args:                                                             │
│ ❱ 1090 │   │   │   raise ValueError(                                                             │
│   1091 │   │   │   │   f"The following `model_kwargs` are not used by the model: {unused_model_  │
│   1092 │   │   │   │   " generate arguments will also show up in this list)"                     │
│   1093 │   │   │   )                                                                             │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
ValueError: The following `model_kwargs` are not used by the model: ['input_ids'] (note: typos in the 
generate arguments will also show up in this list)

Versions:

• Python: 3.10.10
• numpy: 1.24.2
• sentencepiece: 0.1.98
• soundfile: 0.12.1
• tokenizers: 0.13.2
• torch: 2.0.0
• torchaudio: 2.0.1
• transformers: 4.26.1