Adding the rest of wrappers so finetuning for classification is supported out of the box

Adding the rest of the model wrappers (`AutoModelForSequenceClassification`, `AutoModelForMultipleChoice`, `AutoModelForQuestionAnswering` and `AutoModelForTokenClassification`) so finetuning for sequence, token, and other classification tasks is supported out of the box.

config.json

@@ -7,7 +7,11 @@
   "auto_map": {                                                                                                                                                                                                                 
     "AutoConfig": "nomic-ai/nomic-bert-2048--configuration_hf_nomic_bert.NomicBertConfig",
     "AutoModel": "nomic-ai/nomic-bert-2048--modeling_hf_nomic_bert.NomicBertModel",
-    "AutoModelForMaskedLM": "nomic-ai/nomic-bert-2048--modeling_hf_nomic_bert.NomicBertForPreTraining"                                                                                                                                                                                                                         
+    "AutoModelForMaskedLM": "nomic-ai/nomic-bert-2048--modeling_hf_nomic_bert.NomicBertForPreTraining"
+    "AutoModelForSequenceClassification": "nomic-ai/nomic-bert-2048--modeling_hf_nomic_bert.NomicBertForSequenceClassification",
+    "AutoModelForMultipleChoice": "nomic-ai/nomic-bert-2048--modeling_hf_nomic_bert.NomicBertForMultipleChoice",
+    "AutoModelForQuestionAnswering": "nomic-ai/nomic-bert-2048--modeling_hf_nomic_bert.NomicBertForQuestionAnswering",
+    "AutoModelForTokenClassification": "nomic-ai/nomic-bert-2048--modeling_hf_nomic_bert.NomicBertForTokenClassification"
   },
   "bos_token_id": null,
   "causal": false,