feat: support AutoModelForSequenceClassification

modeling_nllbllm2vec.py

@@ -1,12 +1,16 @@
-from typing import Any, Dict, List, Optional, Tuple, cast
+from typing import Any, Dict, List, Optional, Tuple, cast, Union
 
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
 from transformers.models.auto import AutoModel
-from transformers.modeling_outputs import BaseModelOutputWithPooling
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPooling,
+    SequenceClassifierOutputWithPast,
+)
 from transformers.modeling_utils import PreTrainedModel
 from transformers.models.m2m_100.modeling_m2m_100 import M2M100Encoder
+from transformers.cache_utils import Cache
 
 from .configuration_nllbllm2vec import NLLBLLM2VecConfig
 from .modeling_llama_encoder import LlamaEncoderModel
@@ -479,3 +483,94 @@ def repl():
 
     with open("./model.safetensors.index.json", "r") as f:
         print(json.load(f))
+
+
+class NLLBLLM2VecForSequenceClassification(PreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.model = NLLBLLM2Vec(config)
+        self.score = nn.Linear(
+            config.llm2vec_config.hidden_size, self.num_labels, bias=False
+        )
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.model.nllb.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.model.nllb.embed_tokens = value
+
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Union[Cache, List[torch.FloatTensor]]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, SequenceClassifierOutputWithPast]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
+            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        return_dict = (
+            return_dict if return_dict is not None else self.config.use_return_dict
+        )
+
+        transformer_outputs = self.model(
+            input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        hidden_states = transformer_outputs.pooler_output
+        pooled_logits = self.score(hidden_states)
+
+        loss = None
+        if labels is not None:
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (
+                    labels.dtype == torch.long or labels.dtype == torch.int
+                ):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
+                if self.num_labels == 1:
+                    loss = F.mse_loss(pooled_logits.squeeze(), labels.squeeze())
+                else:
+                    loss = F.mse_loss(pooled_logits, labels)
+            elif self.config.problem_type == "single_label_classification":
+                loss = F.cross_entropy(
+                    pooled_logits.view(-1, self.num_labels), labels.view(-1)
+                )
+            elif self.config.problem_type == "multi_label_classification":
+                loss = F.binary_cross_entropy_with_logits(pooled_logits, labels)
+        if not return_dict:
+            output = (pooled_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=pooled_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )