From 0c5d4d87b0badd1b9b6b2d8ac89a453c353a023c Mon Sep 17 00:00:00 2001
From: Meng Zhang <meng@tabbyml.com>
Date: Fri, 15 Sep 2023 10:38:46 +0800
Subject: [PATCH] add placeholder of starcoder in gguf / llama.cpp

---
 gguf-py/gguf/gguf.py | 32 ++++++++++++++++++--------------
 llama.cpp            | 16 +++++++++-------
 2 files changed, 27 insertions(+), 21 deletions(-)

diff --git a/gguf-py/gguf/gguf.py b/gguf-py/gguf/gguf.py
index 7f7204ea1..97792943c 100644
--- a/gguf-py/gguf/gguf.py
+++ b/gguf-py/gguf/gguf.py
@@ -77,13 +77,14 @@ KEY_TOKENIZER_RWKV       = "tokenizer.rwkv.world"
 
 
 class MODEL_ARCH(IntEnum):
-    LLAMA  : int = auto()
-    FALCON : int = auto()
-    BAICHUAN:int = auto()
-    GPT2   : int = auto()
-    GPTJ   : int = auto()
-    GPTNEOX: int = auto()
-    MPT    : int = auto()
+    LLAMA         : int = auto()
+    FALCON        : int = auto()
+    BAICHUAN      : int = auto()
+    GPT2          : int = auto()
+    GPTJ          : int = auto()
+    GPTNEOX       : int = auto()
+    MPT           : int = auto()
+    STARCODER     : int = auto()
 
 
 class MODEL_TENSOR(IntEnum):
@@ -107,13 +108,14 @@ class MODEL_TENSOR(IntEnum):
 
 
 MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
-    MODEL_ARCH.LLAMA:   "llama",
-    MODEL_ARCH.FALCON:  "falcon",
-    MODEL_ARCH.BAICHUAN:"baichuan",
-    MODEL_ARCH.GPT2:    "gpt2",
-    MODEL_ARCH.GPTJ:    "gptj",
-    MODEL_ARCH.GPTNEOX: "gptneox",
-    MODEL_ARCH.MPT:     "mpt",
+    MODEL_ARCH.LLAMA:          "llama",
+    MODEL_ARCH.FALCON:         "falcon",
+    MODEL_ARCH.BAICHUAN:       "baichuan",
+    MODEL_ARCH.GPT2:           "gpt2",
+    MODEL_ARCH.GPTJ:           "gptj",
+    MODEL_ARCH.GPTNEOX:        "gptneox",
+    MODEL_ARCH.MPT:            "mpt",
+    MODEL_ARCH.STARCODER:      "starcoder",
 }
 
 MODEL_TENSOR_NAMES: dict[MODEL_ARCH, dict[MODEL_TENSOR, str]] = {
@@ -171,6 +173,8 @@ MODEL_TENSOR_NAMES: dict[MODEL_ARCH, dict[MODEL_TENSOR, str]] = {
         MODEL_TENSOR.FFN_DOWN:      "blk.{bid}.ffn_down",
         MODEL_TENSOR.FFN_UP:        "blk.{bid}.ffn_up",
     },
+    MODEL_ARCH.STARCODER: {
+    },
     MODEL_ARCH.GPT2: {
         # TODO
     },
diff --git a/llama.cpp b/llama.cpp
index 30728b7cb..2b535a881 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -160,17 +160,19 @@ enum llm_arch {
     LLM_ARCH_GPTJ,
     LLM_ARCH_GPTNEOX,
     LLM_ARCH_MPT,
+    LLM_ARCH_STARCODER,
     LLM_ARCH_UNKNOWN,
 };
 
 static std::map<llm_arch, std::string> LLM_ARCH_NAMES = {
-    { LLM_ARCH_LLAMA,   "llama"   },
-    { LLM_ARCH_FALCON,  "falcon"  },
-    { LLM_ARCH_GPT2,    "gpt2"    },
-    { LLM_ARCH_GPTJ,    "gptj"    },
-    { LLM_ARCH_GPTNEOX, "gptneox" },
-    { LLM_ARCH_MPT,     "mpt"     },
-    { LLM_ARCH_BAICHUAN,"baichuan" },
+    { LLM_ARCH_LLAMA,           "llama"   },
+    { LLM_ARCH_FALCON,          "falcon"  },
+    { LLM_ARCH_GPT2,            "gpt2"    },
+    { LLM_ARCH_GPTJ,            "gptj"    },
+    { LLM_ARCH_GPTNEOX,         "gptneox" },
+    { LLM_ARCH_MPT,             "mpt"     },
+    { LLM_ARCH_BAICHUAN,        "baichuan" },
+    { LLM_ARCH_STARCODER,       "starcoder" },
 };
 
 enum llm_kv {