Add Qwen2Moe GGUF loading support (huggingface#33264)

* update gguf doc, config and tensor mapping

* add qwen2moe architecture support, GGUFQwen2MoeConverter and q4 unit tests

* apply code style fixes

* reformat files

* assign GGUFQwen2Converter to qwen2_moe

docs/source/en/gguf.md

@@ -78,6 +78,7 @@ For now the supported model architectures are the architectures that have been v
 - LLaMa
 - Mistral
 - Qwen2
+- Qwen2Moe
 
 ## Example usage
 

src/transformers/integrations/ggml.py

@@ -79,6 +79,21 @@
         "output.weight": "lm_head.weight",
         "output_norm": "model.norm",
     },
+    "qwen2moe": {
+        "token_embd": "model.embed_tokens",
+        "blk": "model.layers",
+        "ffn_up": "mlp.up_proj",
+        "ffn_down": "mlp.down_proj",
+        "ffn_gate": "mlp.gate_proj",
+        "ffn_norm": "post_attention_layernorm",
+        "attn_norm": "input_layernorm",
+        "attn_q": "self_attn.q_proj",
+        "attn_v": "self_attn.v_proj",
+        "attn_k": "self_attn.k_proj",
+        "attn_output": "self_attn.o_proj",
+        "output.weight": "lm_head.weight",
+        "output_norm": "model.norm",
+    },
 }
 
 
@@ -123,6 +138,18 @@
         "attention.layer_norm_rms_epsilon": "rms_norm_eps",
         "vocab_size": "vocab_size",
     },
+    "qwen2moe": {
+        "context_length": "max_position_embeddings",
+        "block_count": "num_hidden_layers",
+        "feed_forward_length": "intermediate_size",
+        "embedding_length": "hidden_size",
+        "rope.dimension_count": None,
+        "rope.freq_base": "rope_theta",
+        "attention.head_count": "num_attention_heads",
+        "attention.head_count_kv": "num_key_value_heads",
+        "attention.layer_norm_rms_epsilon": "rms_norm_eps",
+        "vocab_size": "vocab_size",
+    },
     "tokenizer": {
         "ggml.bos_token_id": "bos_token_id",
         "ggml.eos_token_id": "eos_token_id",
@@ -244,7 +271,15 @@ def tokenizer(self, proto):
         bos_token = proto.tokens[proto.bos_token_id] if getattr(proto, "bos_token_id", None) is not None else None
         eos_token = proto.tokens[proto.bos_token_id] if getattr(proto, "eos_token_id", None) is not None else None
 
-        tokenizer = Tokenizer(BPE(bpe_vocab, merges, unk_token=unk_token, fuse_unk=True, byte_fallback=True))
+        tokenizer = Tokenizer(
+            BPE(
+                bpe_vocab,
+                merges,
+                unk_token=unk_token,
+                fuse_unk=True,
+                byte_fallback=True,
+            )
+        )
 
         special_tokens = []
 
@@ -358,6 +393,7 @@ def converted(self) -> Tokenizer:
 GGUF_TO_FAST_CONVERTERS = {
     "llama": GGUFLlamaConverter,
     "qwen2": GGUFQwen2Converter,
+    "qwen2_moe": GGUFQwen2Converter,
 }
 
 

src/transformers/modeling_gguf_pytorch_utils.py

@@ -96,6 +96,9 @@ def load_gguf_checkpoint(gguf_checkpoint_path, return_tensors=False):
     else:
         updated_architecture = architecture
 
+    if "qwen2moe" in architecture:
+        updated_architecture = "qwen2_moe"
+
     if architecture not in GGUF_SUPPORTED_ARCHITECTURES:
         raise ValueError(f"Architecture {architecture} not supported")
 

tests/quantization/ggml/test_ggml.py

@@ -16,7 +16,12 @@
 import unittest
 
 from transformers import AddedToken, AutoModelForCausalLM, AutoTokenizer
-from transformers.testing_utils import require_gguf, require_torch_gpu, slow, torch_device
+from transformers.testing_utils import (
+    require_gguf,
+    require_torch_gpu,
+    slow,
+    torch_device,
+)
 from transformers.utils import is_torch_available
 
 
@@ -33,6 +38,7 @@ class GgufIntegrationTests(unittest.TestCase):
     imatrix_model_id = "duyntnet/TinyLlama-1.1B-Chat-v1.0-imatrix-GGUF"
     mistral_model_id = "TheBloke/Mistral-7B-Instruct-v0.2-GGUF"
     qwen2_model_id = "Qwen/Qwen1.5-0.5B-Chat-GGUF"
+    qwen2_moe_model_id = "RichardErkhov/Qwen_-_Qwen1.5-MoE-A2.7B-Chat-gguf"
     llama3_model_id = "NousResearch/Meta-Llama-3-8B-GGUF"
     tinyllama_model_id = "PenutChen/TinyLlama-1.1B-Chat-v1.0-GGUF"
 
@@ -59,6 +65,7 @@ class GgufIntegrationTests(unittest.TestCase):
 
     q4_0_mistral_model_id = "mistral-7b-instruct-v0.2.Q4_0.gguf"
     q4_0_qwen2_model_id = "qwen1_5-0_5b-chat-q4_0.gguf"
+    q4_0_qwen2_moe_model_id = "Qwen1.5-MoE-A2.7B-Chat.Q4_0.gguf"
     q4_llama3_model_id = "Meta-Llama-3-8B-Q4_K_M.gguf"
     f16_tinyllama_model_id = "TinyLlama-1.1B-Chat-v1.0.FP16.gguf"
 
@@ -298,7 +305,10 @@ def test_f16(self):
     def test_mistral_q4_0(self):
         tokenizer = AutoTokenizer.from_pretrained(self.mistral_model_id, gguf_file=self.q4_0_mistral_model_id)
         model = AutoModelForCausalLM.from_pretrained(
-            self.mistral_model_id, gguf_file=self.q4_0_mistral_model_id, device_map="auto", torch_dtype=torch.float16
+            self.mistral_model_id,
+            gguf_file=self.q4_0_mistral_model_id,
+            device_map="auto",
+            torch_dtype=torch.float16,
         )
 
         text = tokenizer(self.example_text, return_tensors="pt").to(torch_device)
@@ -310,7 +320,10 @@ def test_mistral_q4_0(self):
     def test_qwen2_q4_0(self):
         tokenizer = AutoTokenizer.from_pretrained(self.qwen2_model_id, gguf_file=self.q4_0_qwen2_model_id)
         model = AutoModelForCausalLM.from_pretrained(
-            self.qwen2_model_id, gguf_file=self.q4_0_qwen2_model_id, device_map="auto", torch_dtype=torch.float16
+            self.qwen2_model_id,
+            gguf_file=self.q4_0_qwen2_model_id,
+            device_map="auto",
+            torch_dtype=torch.float16,
         )
 
         text = tokenizer(self.example_text, return_tensors="pt").to(torch_device)
@@ -319,6 +332,21 @@ def test_qwen2_q4_0(self):
         EXPECTED_TEXT = "Hello.jsoup\n\nI am a beginner"
         self.assertEqual(tokenizer.decode(out[0], skip_special_tokens=True), EXPECTED_TEXT)
 
+    def test_qwen2_moe_q4_0(self):
+        tokenizer = AutoTokenizer.from_pretrained(self.qwen2_moe_model_id, gguf_file=self.q4_0_qwen2_moe_model_id)
+        model = AutoModelForCausalLM.from_pretrained(
+            self.qwen2_moe_model_id,
+            gguf_file=self.q4_0_qwen2_moe_model_id,
+            device_map="auto",
+            torch_dtype=torch.float16,
+        )
+
+        text = tokenizer(self.example_text, return_tensors="pt").to(torch_device)
+        out = model.generate(**text, max_new_tokens=10)
+
+        EXPECTED_TEXT = "Hello everyone, I'm a newbie here and would like"
+        self.assertEqual(tokenizer.decode(out[0], skip_special_tokens=True), EXPECTED_TEXT)
+
     def test_llama3_q4_0_tokenizer(self):
         tokenizer = AutoTokenizer.from_pretrained(self.llama3_model_id, gguf_file=self.q4_llama3_model_id)
         with tempfile.TemporaryDirectory() as tmpdirname:
@@ -331,7 +359,10 @@ def test_llama3_q4_0_tokenizer(self):
     def test_llama3_q4_0(self):
         tokenizer = AutoTokenizer.from_pretrained(self.llama3_model_id, gguf_file=self.q4_llama3_model_id)
         model = AutoModelForCausalLM.from_pretrained(
-            self.llama3_model_id, gguf_file=self.q4_llama3_model_id, device_map="auto", torch_dtype=torch.float16
+            self.llama3_model_id,
+            gguf_file=self.q4_llama3_model_id,
+            device_map="auto",
+            torch_dtype=torch.float16,
         )
 
         text = tokenizer(self.example_text, return_tensors="pt").to(torch_device)