add the pure fp16 (#5591)

PaddlePaddle · Apr 10, 2023 · c2f7e31 · c2f7e31
1 parent 6d1c163
commit c2f7e31
Showing 1 changed file with 0 additions and 2 deletions.
diff --git a/paddlenlp/transformers/bloom/configuration.py b/paddlenlp/transformers/bloom/configuration.py
@@ -108,7 +108,6 @@ def __init__(
         attention_dropout=0.0,
         attention_softmax_in_fp32=True,
         pretraining_tp=1,  # TP rank used when training with megatron
-        dtype="float16",
         slow_but_exact=False,
         use_recompute=False,
         use_pure_fp16=False,
@@ -133,7 +132,6 @@ def __init__(
 
         self.bos_token_id = bos_token_id
         self.eos_token_id = eos_token_id
-        self.dtype = dtype
         self.slow_but_exact = slow_but_exact
         self.use_recompute = use_recompute
         self.use_pure_fp16 = use_pure_fp16