fix: requirements.txt and newer transformer's compatibility

indextts/gpt/transformers_generation_utils.py

@@ -55,7 +55,6 @@
     AssistedCandidateGeneratorDifferentTokenizers,
     CandidateGenerator,
     PromptLookupCandidateGenerator,
-    _crop_past_key_values,
     _prepare_attention_mask,
     _prepare_token_type_ids,
 )
@@ -104,6 +103,42 @@
 )
 
 
+# Compatibility function for older transformers versions
+def _crop_past_key_values(model, past_key_values, new_cache_size):
+    """
+    Crop past_key_values to match new_cache_size for compatibility with newer transformers versions.
+    """
+    if past_key_values is None:
+        return None
+
+    cropped_past_key_values = []
+    for layer_past in past_key_values:
+        if layer_past is None:
+            cropped_past_key_values.append(None)
+            continue
+
+        # Each layer_past should be a tuple of (key, value) tensors
+        if isinstance(layer_past, (tuple, list)) and len(layer_past) == 2:
+            key, value = layer_past
+            # Crop along the sequence length dimension (usually dim=2)
+            if key is not None and key.size(2) > new_cache_size:
+                cropped_key = key[:, :, :new_cache_size, :]
+            else:
+                cropped_key = key
+
+            if value is not None and value.size(2) > new_cache_size:
+                cropped_value = value[:, :, :new_cache_size, :]
+            else:
+                cropped_value = value
+
+            cropped_past_key_values.append((cropped_key, cropped_value))
+        else:
+            # Fallback for unexpected formats
+            cropped_past_key_values.append(layer_past)
+
+    return tuple(cropped_past_key_values)
+
+
 if TYPE_CHECKING:
     from transformers.modeling_utils import PreTrainedModel
     from transformers.tokenization_utils_base import PreTrainedTokenizerBase
@@ -1002,7 +1037,8 @@ def _get_logits_processor(
                     device=device,
                 )
             )
-        if generation_config.forced_decoder_ids is not None:
+        # Check for forced_decoder_ids attribute compatibility with newer transformers versions
+        if hasattr(generation_config, 'forced_decoder_ids') and generation_config.forced_decoder_ids is not None:
             # TODO (sanchit): move this exception to GenerationConfig.validate() when TF & FLAX are aligned with PT
             raise ValueError(
                 "You have explicitly specified `forced_decoder_ids`. Please remove the `forced_decoder_ids` argument "

indextts/gpt/transformers_gpt2.py

@@ -32,7 +32,7 @@
 
 from indextts.gpt.transformers_generation_utils import GenerationMixin
 from indextts.gpt.transformers_modeling_utils import PreTrainedModel
-from transformers.modeling_utils import SequenceSummary
+from transformers.models.gpt2.modeling_gpt2 import GPT2SequenceSummary as SequenceSummary
 
 from transformers.modeling_attn_mask_utils import _prepare_4d_attention_mask_for_sdpa, _prepare_4d_causal_attention_mask_for_sdpa
 from transformers.modeling_outputs import (

requirements.txt

@@ -24,13 +24,7 @@ sentencepiece
 tqdm
 textstat
 
-pynini==2.1.6; platform_system!="Windows"
-WeTextProcessing>=1.0.3; platform_system!="Windows"
-
-WeTextProcessing; platform_machine != "Darwin"
 wetext; platform_system == "Darwin"
 
 # importlib_resources
-# pynini==2.1.6.post1
-# WeTextProcessing>=1.0.4
-# deepspeed # Use it to accelerate model inference
+# deepspeed # Use it to accelerate model inference