check root architecture for convert hf script

Author: piDack

convert_hf_to_gguf.py

@@ -1476,6 +1476,16 @@ def _try_set_pooling_type(self) -> None:
                 raise NotImplementedError("Only MEAN, CLS, and LAST pooling types supported")
             self.gguf_writer.add_pooling_type(pooling_type)
 
+    def _set_vocab_glmedge(self):
+        from transformers import AutoTokenizer
+        tokenizer = AutoTokenizer.from_pretrained(self.dir_model, trust_remote_code=True)
+        special_vocab = gguf.SpecialVocab(self.dir_model, load_merges=True)
+        special_vocab._set_special_token("eos", tokenizer.get_added_vocab()["<|endoftext|>"])
+        special_vocab._set_special_token("eot", tokenizer.get_added_vocab()["<|user|>"])
+        special_vocab._set_special_token("unk", tokenizer.get_added_vocab()["<|endoftext|>"])
+        special_vocab._set_special_token("bos", tokenizer.get_added_vocab()["<|endoftext|>"])
+        special_vocab.add_to_gguf(self.gguf_writer)
+
     def _set_vocab_interns1(self):
         tokens: list[str] = []
         toktypes: list[int] = []
@@ -2359,6 +2369,7 @@ def prepare_tensors(self):
     "VLlama3ForCausalLM",
     "LlavaForConditionalGeneration",
     "VoxtralForConditionalGeneration",
+    "GlmasrModel",
     "LlamaModel")
 class LlamaModel(TextModel):
     model_arch = gguf.MODEL_ARCH.LLAMA
@@ -2410,6 +2421,8 @@ def set_vocab(self):
         # Apply to granite small models only
         if self.hparams.get("vocab_size", 32000) == 49152:
             self.gguf_writer.add_add_bos_token(False)
+        if self.hf_arch == "GlmasrModel":
+            self._set_vocab_glmedge()
 
     def set_gguf_parameters(self):
         super().set_gguf_parameters()
@@ -2564,22 +2577,6 @@ def set_gguf_parameters(self):
             self.gguf_writer.add_rope_scaling_orig_ctx_len(rope_scaling["original_max_position_embeddings"])
 
 
-@ModelBase.register("GlmasrModel")
-class GlmasrModel(LlamaModel):
-    model_arch = gguf.MODEL_ARCH.LLAMA
-
-    def set_vocab(self):
-        super().set_vocab()
-        from transformers import AutoTokenizer
-        tokenizer = AutoTokenizer.from_pretrained(self.dir_model, trust_remote_code=True)
-        special_vocab = gguf.SpecialVocab(self.dir_model, load_merges=True)
-        special_vocab._set_special_token("eos", tokenizer.get_added_vocab()["<|endoftext|>"])
-        special_vocab._set_special_token("eot", tokenizer.get_added_vocab()["<|user|>"])
-        special_vocab._set_special_token("unk", tokenizer.get_added_vocab()["<|endoftext|>"])
-        special_vocab._set_special_token("bos", tokenizer.get_added_vocab()["<|endoftext|>"])
-        special_vocab.add_to_gguf(self.gguf_writer)
-
-
 @ModelBase.register("AfmoeForCausalLM")
 class AfmoeModel(LlamaModel):
     model_arch = gguf.MODEL_ARCH.AFMOE