From 2a9498d5df6319fc294752f4574c0a23727c84c4 Mon Sep 17 00:00:00 2001
From: jeromeku <jerome.ku@gmail.com>
Date: Mon, 31 Mar 2025 11:23:22 -0700
Subject: [PATCH] add deepseek v3

---
 unsloth/registry/_deepseek.py | 53 +++++++++++++++++++++++++++++++++++
 unsloth/registry/registry.py  |  3 ++
 2 files changed, 56 insertions(+)
 create mode 100644 unsloth/registry/_deepseek.py

diff --git a/unsloth/registry/_deepseek.py b/unsloth/registry/_deepseek.py
new file mode 100644
index 0000000000..8bdcd3c2e7
--- /dev/null
+++ b/unsloth/registry/_deepseek.py
@@ -0,0 +1,53 @@
+from unsloth.registry.registry import ModelInfo, ModelMeta, QuantType, _register_models
+
+_IS_DEEPSEEKV3_REGISTERED = False
+
+class DeepseekV3ModelInfo(ModelInfo):
+    @classmethod
+    def construct_model_name(cls, base_name, version, size, quant_type, instruct_tag):
+        key = f"{base_name}-V{version}"
+        key = cls.append_instruct_tag(key, instruct_tag)
+        key = cls.append_quant_type(key, quant_type)
+        return key
+
+# Deepseek V3 Model Meta
+DeepseekV3Meta = ModelMeta(
+    org="deepseek-ai",
+    base_name="DeepSeek",
+    instruct_tags=[None],
+    model_version="3",
+    model_sizes=[""],
+    model_info_cls=DeepseekV3ModelInfo,
+    is_multimodal=False,
+    quant_types=[QuantType.NONE, QuantType.BF16],
+)
+
+DeepseekV3_0324Meta = ModelMeta(
+    org="deepseek-ai",
+    base_name="DeepSeek",
+    instruct_tags=[None],
+    model_version="3-0324",
+    model_sizes=[""],
+    model_info_cls=DeepseekV3ModelInfo,
+    is_multimodal=False,
+    quant_types=[QuantType.NONE, QuantType.GGUF],
+)
+
+def register_deepseek_v3_models(include_original_model: bool = False):
+    global _IS_DEEPSEEKV3_REGISTERED
+    if _IS_DEEPSEEKV3_REGISTERED:
+        return
+    _register_models(DeepseekV3Meta, include_original_model=include_original_model)
+    _register_models(DeepseekV3_0324Meta, include_original_model=include_original_model)
+    _IS_DEEPSEEKV3_REGISTERED = True
+
+register_deepseek_v3_models(include_original_model=True)
+
+if __name__ == "__main__":
+    from unsloth.registry.registry import MODEL_REGISTRY, _check_model_info
+    for model_id, model_info in MODEL_REGISTRY.items():
+        model_info = _check_model_info(model_id)
+        if model_info is None:
+            print(f"\u2718 {model_id}")
+        else:
+            print(f"\u2713 {model_id}")
diff --git a/unsloth/registry/registry.py b/unsloth/registry/registry.py
index 869a7efb5d..1eee884259 100644
--- a/unsloth/registry/registry.py
+++ b/unsloth/registry/registry.py
@@ -8,17 +8,20 @@ class QuantType(Enum):
     UNSLOTH = "unsloth" # dynamic 4-bit quantization
     GGUF = "GGUF"
     NONE = "none"
+    BF16 = "bf16" # only for Deepseek V3
 
 # Tags for Hugging Face model paths
 BNB_QUANTIZED_TAG = "bnb-4bit"
 UNSLOTH_DYNAMIC_QUANT_TAG = "unsloth" + "-" + BNB_QUANTIZED_TAG
 GGUF_TAG = "GGUF"
+BF16_TAG = "bf16"
 
 QUANT_TAG_MAP = {
     QuantType.BNB: BNB_QUANTIZED_TAG,
     QuantType.UNSLOTH: UNSLOTH_DYNAMIC_QUANT_TAG,
     QuantType.GGUF: GGUF_TAG,
     QuantType.NONE: None,
+    QuantType.BF16: BF16_TAG,
 } 
 
 # NOTE: models registered with org="unsloth" and QUANT_TYPE.NONE are aliases of QUANT_TYPE.UNSLOTH