create_model.py

from transformers import SpeechEncoderDecoderModel, AutoFeatureExtractor, GPT2Tokenizer
import torch

# checkpoints to leverage
encoder_id = "facebook/wav2vec2-large-lv60"
decoder_id = "gpt2-medium"

model = SpeechEncoderDecoderModel.from_encoder_decoder_pretrained(encoder_id, decoder_id, encoder_add_adapter=False)
model.config.encoder.feat_proj_dropout = 0.0
model.config.encoder.final_dropout = 0.0
model.config.encoder.mask_time_prob = 0.1


# force GPT2 to append EOS to begin and end of seq
def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None):
    outputs = [self.bos_token_id] + token_ids_0 + [self.eos_token_id]
    return outputs


GPT2Tokenizer.build_inputs_with_special_tokens = build_inputs_with_special_tokens
gpt2_tokenizer = GPT2Tokenizer.from_pretrained(decoder_id)
# set pad_token_id to unk_token_id, note: unk_token_id == eos_token_id == bos_token_id
gpt2_tokenizer.pad_token = gpt2_tokenizer.unk_token
gpt2_tokenizer.save_pretrained("./")

model.config.pad_token_id = gpt2_tokenizer.pad_token_id
model.config.decoder_start_token_id = model.decoder.config.bos_token_id
model.config.eos_token_id = model.decoder.config.eos_token_id
model.config.max_length = 50
model.config.num_beams = 1
model.config.encoder.layerdrop = 0.0
model.config.use_cache = False
model.config.decoder.use_cache = False
model.config.processor_class = "Wav2Vec2Processor"

# check if generation works
out = model.generate(torch.ones((1, 2000)))

model.save_pretrained("./")

feature_extractor = AutoFeatureExtractor.from_pretrained(encoder_id)
feature_extractor.save_pretrained("./")