You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
zhanghangit a2e70f02e6 update code and readme 2 years ago
..
.ipynb_checkpoints update code and readme 2 years ago
__pycache__ update code and readme 2 years ago
dynamicconv_layer update code and readme 2 years ago
lightconv_layer update code and readme 2 years ago
quantization update code and readme 2 years ago
__init__.py update code and readme 2 years ago
adaptive_input.py update code and readme 2 years ago
adaptive_softmax.py update code and readme 2 years ago
base_layer.py update code and readme 2 years ago
beamable_mm.py update code and readme 2 years ago
character_token_embedder.py update code and readme 2 years ago
checkpoint_activations.py update code and readme 2 years ago
conv_tbc.py update code and readme 2 years ago
cross_entropy.py update code and readme 2 years ago
cuda_utils.cu update code and readme 2 years ago
downsampled_multihead_attention.py update code and readme 2 years ago
dynamic_convolution.py update code and readme 2 years ago
dynamic_crf_layer.py update code and readme 2 years ago
fairseq_dropout.py update code and readme 2 years ago
fp32_group_norm.py update code and readme 2 years ago
gelu.py update code and readme 2 years ago
grad_multiply.py update code and readme 2 years ago
gumbel_vector_quantizer.py update code and readme 2 years ago
kmeans_vector_quantizer.py update code and readme 2 years ago
layer_drop.py update code and readme 2 years ago
layer_norm.py update code and readme 2 years ago
learned_positional_embedding.py update code and readme 2 years ago
lightweight_convolution.py update code and readme 2 years ago
linearized_convolution.py update code and readme 2 years ago
multihead_attention.py update code and readme 2 years ago
positional_embedding.py update code and readme 2 years ago
quant_noise.py update code and readme 2 years ago
same_pad.py update code and readme 2 years ago
scalar_bias.py update code and readme 2 years ago
sinusoidal_positional_embedding.py update code and readme 2 years ago
sparse_multihead_attention.py update code and readme 2 years ago
sparse_transformer_sentence_encoder.py update code and readme 2 years ago
sparse_transformer_sentence_encoder_layer.py update code and readme 2 years ago
transformer_layer.py update code and readme 2 years ago
transformer_sentence_encoder.py update code and readme 2 years ago
transformer_sentence_encoder_layer.py update code and readme 2 years ago
transpose_last.py update code and readme 2 years ago
unfold.py update code and readme 2 years ago
vggblock.py update code and readme 2 years ago

鹏程-通言模型 通言模型是在M2M-100模型结构上进行改进的多语种机器翻译模型,通过参数复用和增量式训练,将模型参数从1.2B提升至13.2B,在一带一路多个小语种的翻译上大幅提升。

Text Python C++ Cuda other