Fix stuff related to the causal_mask in CodeGen. (#21527)

* Fix stuff related to the causal_mask in CodeGen.

1. Line 613, `_keys_to_ignore_on_load_missing  =  [r"h\.\d+\.attn\.masked_bias", r"h\.\d+\.attn\.bias"]` => `_keys_to_ignore_on_load_missing  =  [r"h\.\d+\.attn\.causal_mask"]` to load correctly from CodeGen checkpoint without `causal_mask`.
2. Line 152, `causal_mask = self.causal_mask[:, :, key_length - query_length : key_length, :key_length]
` => `causal_mask = self.causal_mask[:, :, key_length - query_length : key_length, :key_length].bool()
` to alleviate potential user warning saying like `UserWarning: where received a uint8 condition tensor. This behavior is deprecated and will be removed in a future version of PyTorch. Use a boolean condition instead.`.

* Revert the .bool()

Revert the .bool() and leave it to the future PR.

src/transformers/models/codegen/modeling_codegen.py

@@ -610,7 +610,7 @@ class CodeGenModel(CodeGenPreTrainedModel):
     CODEGEN_START_DOCSTRING,
 )
 class CodeGenForCausalLM(CodeGenPreTrainedModel):
-    _keys_to_ignore_on_load_missing = [r"h\.\d+\.attn\.masked_bias", r"h\.\d+\.attn\.bias"]
+    _keys_to_ignore_on_load_missing = [r"h\.\d+\.attn\.causal_mask"]
 
     def __init__(self, config):
         super().__init__(config)