Upload folder using huggingface_hub

modeling_pantagruel_uni.py

@@ -23,6 +23,7 @@
 
 """ PantagruelUni model."""
 import math
+import logging
 import warnings
 from typing import Optional, Tuple, Dict, List, Callable, Any, Union
 from functools import partial
@@ -40,10 +41,7 @@ from transformers import PreTrainedModel
 from transformers.utils import (
     ModelOutput, TransformersKwargs, auto_docstring
 )
-from transformers.activations import ACT2FN, gelu
-from transformers.modeling_attn_mask_utils import (
-    _prepare_4d_attention_mask, _prepare_4d_attention_mask_for_sdpa
-)
+from transformers.activations import gelu
 from transformers.utils.generic import can_return_tuple
 from transformers.processing_utils import Unpack
 from transformers.modeling_outputs import (
@@ -55,7 +53,7 @@ from transformers.modeling_outputs import (
     CausalLMOutput,
     XVectorOutput,
 )
-from transformers.utils import auto_docstring, is_peft_available
+from transformers.utils import is_peft_available
 from .configuration_pantagruel_uni import (
     PantagruelUniConfig,
     PantagruelModalityConfig,
@@ -1769,7 +1767,7 @@ class PantagruelUniForMaskedLM(PantagruelUniPreTrainedModel):
         super().__init__(config)
 
         if config.is_decoder:
-            logger.warning(
+            logging.warning(
                 "If you want to use `PantagruelTextForMaskedLM` make sure `config.is_decoder=False` for "
                 "bi-directional self-attention."
             )
@@ -1845,10 +1843,10 @@ class PantagruelUniForSequenceClassification(PantagruelUniPreTrainedModel):
         self.pantagruel_uni = PantagruelUniModel(config, add_pooling_layer=False)
 
         if config.supported_modality == "TEXT":
-            logger.info("Initializing PantagruelUniForSequenceClassification for TEXT")
+            logging.info("Initializing PantagruelUniForSequenceClassification for TEXT")
             self.classifier = PantagruelTextClassificationHead(config)
         elif config.supported_modality == "AUDIO":
-            logger.info("Initializing PantagruelUniForSequenceClassification for AUDIO")
+            logging.info("Initializing PantagruelUniForSequenceClassification for AUDIO")
             num_layers = config.num_hidden_layers + 1  # transformer layers + input embeddings
             if config.modalities.audio.use_weighted_layer_sum:
                 self.layer_weights = nn.Parameter(torch.ones(num_layers) / num_layers)