TEMPORARY llama fix

src/nnsight/models/LanguageModel.py

@@ -4,9 +4,15 @@
 from typing import Any, Dict, List, Optional, Tuple, Type, Union
 
 import torch
-from transformers import (AutoConfig, AutoModel, AutoModelForCausalLM,
-                          AutoTokenizer, BatchEncoding, PreTrainedModel,
-                          PreTrainedTokenizer)
+from transformers import (
+    AutoConfig,
+    AutoModel,
+    AutoModelForCausalLM,
+    AutoTokenizer,
+    BatchEncoding,
+    PreTrainedModel,
+    PreTrainedTokenizer,
+)
 from transformers.models.auto import modeling_auto
 from transformers.models.llama.configuration_llama import LlamaConfig
 
@@ -139,21 +145,20 @@ def __init__(
             if not isinstance(automodel, str)
             else getattr(modeling_auto, automodel)
         )
-        
+
         if isinstance(model_key, torch.nn.Module):
-            
-            setattr(model_key, 'generator', WrapperModule())
+
+            setattr(model_key, "generator", WrapperModule())
 
         super().__init__(model_key, *args, **kwargs)
 
     def _load(
         self, repo_id: str, tokenizer_kwargs: Optional[Dict[str, Any]] = None, **kwargs
     ) -> PreTrainedModel:
 
-        config = kwargs.pop("config", None) or AutoConfig.from_pretrained(repo_id, **kwargs)
-
-
-
+        config = kwargs.pop("config", None) or AutoConfig.from_pretrained(
+            repo_id, **kwargs
+        )
 
         if self.tokenizer is None:
             if tokenizer_kwargs is None:
@@ -163,31 +168,29 @@ def _load(
                 warnings.warn(
                     "NNsight LanguageModel requires padding_side='left' for tokenizers, setting it to 'left'"
                 )
-                
+
             self.tokenizer = AutoTokenizer.from_pretrained(
                 repo_id, config=config, padding_side="left", **tokenizer_kwargs
             )
             self.tokenizer.pad_token = self.tokenizer.eos_token
 
         if self._model is None:
-
-
-            if isinstance(config, LlamaConfig) and  "rope_type" in config.rope_scaling:
-                config.rope_scaling['rope_type'] = "default"
+
+            if isinstance(config, LlamaConfig) and isinstance(config.rope_scaling, dict) and "rope_type" in config.rope_scaling:
+                config.rope_scaling["rope_type"] = "default"
 
             model = self.automodel.from_config(config, trust_remote_code=True)
-            
-            setattr(model, 'generator', WrapperModule())
+
+            setattr(model, "generator", WrapperModule())
 
             return model
-
-        if isinstance(config, LlamaConfig) and  "rope_type" in config.rope_scaling:
-            config.rope_scaling['rope_type'] = "llama3"
 
+        if isinstance(config, LlamaConfig) and isinstance(config.rope_scaling, dict) and "rope_type" in config.rope_scaling:
+            config.rope_scaling["rope_type"] = "llama3"
 
         model = self.automodel.from_pretrained(repo_id, config=config, **kwargs)
 
-        setattr(model, 'generator', WrapperModule())
+        setattr(model, "generator", WrapperModule())
 
         return model