Build the model.
(args)
| 13 | |
| 14 | |
| 15 | def model_provider(args): |
| 16 | """Build the model.""" |
| 17 | |
| 18 | old_dtype = paddle.get_default_dtype() |
| 19 | paddle.set_default_dtype("float16") |
| 20 | model = CodeGeeXModel( |
| 21 | args.hidden_size, |
| 22 | args.num_layers, |
| 23 | args.num_attention_heads, |
| 24 | args.padded_vocab_size, |
| 25 | args.max_position_embeddings |
| 26 | ) |
| 27 | model.language_model.embedding.word_embeddings.to(dtype="float32") |
| 28 | model.language_model.embedding.position_embeddings.to(dtype="float32") |
| 29 | model.language_model.topQueryEmbedding.top_query_embeddings.to(dtype="float32") |
| 30 | for i in model.language_model.transformer.layers: |
| 31 | i.input_layernorm.to(dtype="float32") |
| 32 | i.post_attention_layernorm.to(dtype="float32") |
| 33 | model.language_model.transformer.topQueryLayer.input_layernorm.to(dtype="float32") |
| 34 | model.language_model.transformer.topQueryLayer.post_attention_layernorm.to(dtype="float32") |
| 35 | model.language_model.transformer.final_layernorm.to(dtype="float32") |
| 36 | paddle.set_default_dtype(old_dtype) |
| 37 | |
| 38 | return model |
| 39 | |
| 40 | |
| 41 | def add_code_generation_args(parser): |