Add support for GGUF Phi-3 (#31844)

* Update docs for GGUF supported models

* Add tensor mappings and define class GGUFPhi3Converter

* Fix tokenizer

* Working version

* Attempt to fix some CI failures

* Run ruff format

* Add vocab, merges, decoder methods like LlamaConverter

* Resolve conflicts since Qwen2Moe was added to gguf

- I missed one place when resolving conflict
- I also made a mistake with tests_ggml.py and now has been fixed to reflect
its master version.

tests/quantization/ggml/test_ggml.py

@@ -41,6 +41,7 @@ class GgufIntegrationTests(unittest.TestCase):
     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"
+    phi3_model_id = "microsoft/Phi-3-mini-4k-instruct-gguf"
 
     # standard quants
     q4_0_gguf_model_id = "tinyllama-1.1b-chat-v1.0.Q4_0.gguf"
@@ -63,6 +64,7 @@ class GgufIntegrationTests(unittest.TestCase):
     iq4_xs_gguf_model_id = "TinyLlama-1.1B-Chat-v1.0-IQ4_XS.gguf"
     iq4_nl_gguf_model_id = "TinyLlama-1.1B-Chat-v1.0-IQ4_NL.gguf"
 
+    q4_0_phi3_model_id = "Phi-3-mini-4k-instruct-q4.gguf"
     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"
@@ -347,6 +349,18 @@ class GgufIntegrationTests(unittest.TestCase):
         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_phi3_q4_0(self):
+        tokenizer = AutoTokenizer.from_pretrained(self.phi3_model_id, gguf_file=self.q4_0_phi3_model_id)
+        model = AutoModelForCausalLM.from_pretrained(
+            self.phi3_model_id, gguf_file=self.q4_0_phi3_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, I've been reading about the impact of"
+        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: