Upload JointCTCAttentionEncoderDecoder

config.json

@@ -5,7 +5,7 @@
   ],
   "auto_map": {
     "AutoConfig": "configuration_decred.JointCTCAttentionEncoderDecoderConfig",
-    "AutoModelForSpeechSeq2Seq": "modeling_decred.JointCTCAttentionEncoderDecoder"
+    "AutoModel": "modeling_decred.JointCTCAttentionEncoderDecoder"
   },
   "ctc_weight": 0.3,
   "decoder": {

generation.py

@@ -0,0 +1,61 @@
+from transformers import GenerationConfig
+
+
+class GenerationConfigCustom(GenerationConfig):
+    def __init__(
+        self,
+        ctc_weight=0.0,
+        ctc_margin=0,
+        lm_weight=0,
+        lm_model=None,
+        space_token_id=-1,
+        eos_space_trick_weight=0,
+        apply_eos_space_trick=False,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+        self.ctc_weight = ctc_weight
+        self.ctc_margin = ctc_margin
+        self.lm_weight = lm_weight
+        self.lm_model = lm_model
+        self.space_token_id = space_token_id
+        self.eos_space_trick_weight = eos_space_trick_weight
+        self.apply_eos_space_trick = apply_eos_space_trick
+
+    def update_from_string(self, update_str: str):
+        """
+        Updates attributes of this class with attributes from `update_str`.
+
+        The expected format is ints, floats and strings as is, and for booleans use `true` or `false`. For example:
+        "n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
+
+        The keys to change have to already exist in the config object.
+
+        Args:
+            update_str (`str`): String with attributes that should be updated for this class.
+
+        """
+
+        d = dict(x.split("=") for x in update_str.split(";"))
+        for k, v in d.items():
+            if not hasattr(self, k):
+                raise ValueError(f"key {k} isn't in the original config dict")
+
+            old_v = getattr(self, k)
+            if isinstance(old_v, bool):
+                if v.lower() in ["true", "1", "y", "yes"]:
+                    v = True
+                elif v.lower() in ["false", "0", "n", "no"]:
+                    v = False
+                else:
+                    raise ValueError(f"can't derive true or false from {v} (key {k})")
+            elif isinstance(old_v, int):
+                v = int(v)
+            elif isinstance(old_v, float):
+                v = float(v)
+            elif not isinstance(old_v, str):
+                raise ValueError(
+                    f"You can only update int, float, bool or string values in the config, got {v} for key {k}"
+                )
+
+            setattr(self, k, v)

modeling_decred.py

@@ -8,7 +8,6 @@ from transformers import (
     AutoConfig,
     AutoModelForCausalLM,
     AutoModelForSpeechSeq2Seq,
-    GenerationConfig,
     LogitsProcessor,
     PretrainedConfig,
     PreTrainedModel,
@@ -28,6 +27,7 @@ from .auto_wrappers import CustomAutoModelForCTC
 from .configuration_decred import JointCTCAttentionEncoderDecoderConfig
 from .ctc_scorer import CTCRescorerLogitsProcessor, LogSoftmaxProcessor
 from .embeddings import AdaptiveEmbedding, PositionalEmbedding
+from .generation import GenerationConfigCustom
 from .multi_head_gpt2 import GPT2LMMultiHeadModel
 
 logger = logging.get_logger("transformers")
@@ -433,7 +433,7 @@ class JointCTCAttentionEncoderDecoder(SpeechEncoderDecoderModel):
 
     def _get_logits_processor(
         self,
-        generation_config: GenerationConfig,
+        generation_config: GenerationConfigCustom,
         input_ids_seq_length: int,
         encoder_input_ids: torch.LongTensor,
         prefix_allowed_tokens_fn: Callable[[int, torch.Tensor], List[int]],
@@ -464,9 +464,13 @@ class JointCTCAttentionEncoderDecoder(SpeechEncoderDecoderModel):
                 self.generation_config.ctc_margin,
                 self.generation_config.ctc_weight,
                 self.generation_config.num_beams,
-                self.generation_config.space_token_id,
-                self.generation_config.apply_eos_space_trick,
-                self.generation_config.eos_space_trick_weight,
+                self.generation_config.space_token_id if hasattr(self.generation_config, "space_token_id") else None,
+                self.generation_config.apply_eos_space_trick
+                if hasattr(self.generation_config, "apply_eos_space_trick")
+                else False,
+                self.generation_config.eos_space_trick_weight
+                if hasattr(self.generation_config, "eos_space_trick_weight")
+                else 0.0,
             )
             processors.append(self.ctc_rescorer)
         if hasattr(generation_config, "lm_weight") and generation_config.lm_weight > 0:
@@ -524,7 +528,7 @@ class JointCTCAttentionEncoderDecoder(SpeechEncoderDecoderModel):
     def generate(
         self,
         inputs: Optional[torch.Tensor] = None,
-        generation_config: Optional[GenerationConfig] = None,
+        generation_config: Optional[GenerationConfigCustom] = None,
         logits_processor: Optional[LogitsProcessorList] = None,
         stopping_criteria: Optional[StoppingCriteriaList] = None,
         prefix_allowed_tokens_fn: Optional[Callable[[int, torch.Tensor], List[int]]] = None,