llama: allow to override model rope type

Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
ggerganov · May 24, 2024 · 0211330 · 0211330
1 parent 120f7bf
commit 0211330
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 3 deletions.
diff --git a/gguf-py/gguf/constants.py b/gguf-py/gguf/constants.py
@@ -57,6 +57,7 @@ class Attention:
         CAUSAL            = "{arch}.attention.causal"
 
     class Rope:
+        TYPE                    = "{arch}.rope.type"
         DIMENSION_COUNT         = "{arch}.rope.dimension_count"
         FREQ_BASE               = "{arch}.rope.freq_base"
         SCALING_TYPE            = "{arch}.rope.scaling.type"
@@ -806,6 +807,13 @@ class TokenType(IntEnum):
     BYTE         = 6
 
 
+class RopeType(Enum):
+    NONE = 'none'
+    NORM = 'norm'
+    NEOX = 'neox'
+    GLM  = 'glm'
+
+
 class RopeScalingType(Enum):
     NONE   = 'none'
     LINEAR = 'linear'
@@ -998,6 +1006,7 @@ def get_type(val: Any) -> GGUFValueType:
 KEY_ATTENTION_LAYERNORM_RMS_EPS = Keys.Attention.LAYERNORM_RMS_EPS
 
 # RoPE
+KEY_ROPE_TYPE                 = Keys.Rope.TYPE
 KEY_ROPE_DIMENSION_COUNT      = Keys.Rope.DIMENSION_COUNT
 KEY_ROPE_FREQ_BASE            = Keys.Rope.FREQ_BASE
 KEY_ROPE_SCALING_TYPE         = Keys.Rope.SCALING_TYPE

diff --git a/gguf-py/gguf/gguf_writer.py b/gguf-py/gguf/gguf_writer.py
@@ -427,6 +427,9 @@ def add_rope_dimension_count(self, count: int) -> None:
     def add_rope_freq_base(self, value: float) -> None:
         self.add_float32(Keys.Rope.FREQ_BASE.format(arch=self.arch), value)
 
+    def add_rope_type(self, value: RopeType) -> None:
+        self.add_string(Keys.Rope.TYPE.format(arch=self.arch), value.value)
+
     def add_rope_scaling_type(self, value: RopeScalingType) -> None:
         self.add_string(Keys.Rope.SCALING_TYPE.format(arch=self.arch), value.value)
 

diff --git a/llama.cpp b/llama.cpp
@@ -297,6 +297,7 @@ enum llm_kv {
     LLM_KV_ATTENTION_LAYERNORM_RMS_EPS,
     LLM_KV_ATTENTION_CAUSAL,
 
+    LLM_KV_ROPE_TYPE,
     LLM_KV_ROPE_DIMENSION_COUNT,
     LLM_KV_ROPE_FREQ_BASE,
     LLM_KV_ROPE_SCALE_LINEAR,
@@ -375,6 +376,7 @@ static const std::map<llm_kv, const char *> LLM_KV_NAMES = {
     { LLM_KV_ATTENTION_LAYERNORM_RMS_EPS,   "%s.attention.layer_norm_rms_epsilon" },
     { LLM_KV_ATTENTION_CAUSAL,              "%s.attention.causal"                 },
 
+    { LLM_KV_ROPE_TYPE,                     "%s.rope.type"                            },
     { LLM_KV_ROPE_DIMENSION_COUNT,          "%s.rope.dimension_count"                 },
     { LLM_KV_ROPE_FREQ_BASE,                "%s.rope.freq_base"                       },
     { LLM_KV_ROPE_SCALE_LINEAR,             "%s.rope.scale_linear"                    },
@@ -1129,12 +1131,29 @@ struct LLM_TN {
 // gguf helpers
 //
 
+static const std::map<enum llama_rope_type, const char *> LLAMA_ROPE_TYPES = {
+    { LLAMA_ROPE_TYPE_NONE, "none" },
+    { LLAMA_ROPE_TYPE_NORM, "norm" },
+    { LLAMA_ROPE_TYPE_NEOX, "neox" },
+    { LLAMA_ROPE_TYPE_GLM,  "glm"  },
+};
+
 static const std::map<llama_rope_scaling_type, const char *> LLAMA_ROPE_SCALING_TYPES = {
     { LLAMA_ROPE_SCALING_TYPE_NONE,   "none"   },
     { LLAMA_ROPE_SCALING_TYPE_LINEAR, "linear" },
     { LLAMA_ROPE_SCALING_TYPE_YARN,   "yarn"   },
 };
 
+static enum llama_rope_type llama_rope_type_from_string(const std::string & name) {
+    for (const auto & kv : LLAMA_ROPE_TYPES) {
+        if (kv.second == name) {
+            return (enum llama_rope_type) kv.first;
+        }
+    }
+
+    return LLAMA_ROPE_TYPE_NONE;
+}
+
 static llama_rope_scaling_type llama_rope_scaling_type_from_string(const std::string & name) {
     for (const auto & kv : LLAMA_ROPE_SCALING_TYPES) {
         if (kv.second == name) {
@@ -4394,7 +4413,15 @@ static void llm_load_hparams(
         hparams.use_alibi = true;
     }
 
-    hparams.rope_type = llama_rope_type(&model);
+    hparams.rope_type = llama_default_rope_type(&model);
+
+    const auto kv = LLM_KV(model.arch);
+    const int rope_type_keyidx = gguf_find_key(ctx, kv(LLM_KV_ROPE_TYPE).c_str());
+    if (rope_type_keyidx != -1) {
+        std::string rope_type("none");
+        ml.get_key(LLM_KV_ROPE_TYPE, rope_type);
+        hparams.rope_type =  llama_rope_type_from_string(rope_type);
+    }
 }
 
 // TODO: This should probably be in llama.h
@@ -16216,7 +16243,7 @@ enum llama_vocab_type llama_vocab_type(const struct llama_model * model) {
     return model->vocab.type;
 }
 
-enum llama_rope_type llama_rope_type(const struct llama_model * model) {
+enum llama_rope_type llama_default_rope_type(const struct llama_model * model) {
     switch (model->arch) {
         // these models do not use RoPE
         case LLM_ARCH_GPT2:

diff --git a/llama.h b/llama.h
@@ -422,7 +422,7 @@ extern "C" {
     LLAMA_API enum llama_pooling_type llama_pooling_type(const struct llama_context * ctx);
 
     LLAMA_API enum llama_vocab_type   llama_vocab_type  (const struct llama_model   * model);
-    LLAMA_API enum llama_rope_type    llama_rope_type   (const struct llama_model   * model);
+    LLAMA_API enum llama_rope_type    llama_default_rope_type   (const struct llama_model   * model);
 
     LLAMA_API int32_t llama_n_vocab    (const struct llama_model * model);
     LLAMA_API int32_t llama_n_ctx_train(const struct llama_model * model);