diff --git a/convert-hf-to-gguf.py b/convert-hf-to-gguf.py
index 99c1fdb4..63d50f8f 100755
--- a/convert-hf-to-gguf.py
+++ b/convert-hf-to-gguf.py
@@ -1325,8 +1325,6 @@ class LlamaModel(Model):
         # Apply to granite small models only
         if self.hparams.get("vocab_size", 32000) == 49152:
             self.gguf_writer.add_add_bos_token(False)
-            self.gguf_writer.add_rope_type(gguf.RopeType.NEOX)
-            self.gguf_writer.add_rope_scaling_type(gguf.RopeScalingType.NONE)
 
     @staticmethod
     def permute(weights: Tensor, n_head: int, n_head_kv: int | None):
@@ -1342,12 +1340,10 @@ class LlamaModel(Model):
         n_head = self.hparams["num_attention_heads"]
         n_kv_head = self.hparams.get("num_key_value_heads")
 
-        # Skip for granite models
-        if self.hparams.get("vocab_size", 32000) != 49152:
-            if name.endswith("q_proj.weight"):
-                data_torch = LlamaModel.permute(data_torch, n_head, n_head)
-            if name.endswith("k_proj.weight"):
-                data_torch = LlamaModel.permute(data_torch, n_head, n_kv_head)
+        if name.endswith(("q_proj.weight", "q_proj.bias")):
+            data_torch = LlamaModel.permute(data_torch, n_head, n_head)
+        if name.endswith(("k_proj.weight", "k_proj.bias")):
+            data_torch = LlamaModel.permute(data_torch, n_head, n_kv_head)
 
         # process the experts separately
         if name.find("block_sparse_moe.experts") != -1:
diff --git a/gguf-py/gguf/constants.py b/gguf-py/gguf/constants.py
index d5c3d7b5..c9ae259e 100644
--- a/gguf-py/gguf/constants.py
+++ b/gguf-py/gguf/constants.py
@@ -57,7 +57,6 @@ class Keys:
         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"
@@ -807,13 +806,6 @@ class TokenType(IntEnum):
     BYTE         = 6
 
 
-class RopeType(Enum):
-    NONE = 'none'
-    NORM = 'norm'
-    NEOX = 'neox'
-    GLM  = 'glm'
-
-
 class RopeScalingType(Enum):
     NONE   = 'none'
     LINEAR = 'linear'
@@ -1006,7 +998,6 @@ KEY_ATTENTION_LAYERNORM_EPS     = Keys.Attention.LAYERNORM_EPS
 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
index ebfd15fd..8b41b54e 100644
--- a/gguf-py/gguf/gguf_writer.py
+++ b/gguf-py/gguf/gguf_writer.py
@@ -427,9 +427,6 @@ class GGUFWriter:
     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
index 16c11d43..f970c175 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -297,7 +297,6 @@ 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,
@@ -376,7 +375,6 @@ 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"                    },
@@ -1131,29 +1129,12 @@ 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) {
@@ -4417,15 +4398,7 @@ static void llm_load_hparams(
         hparams.use_alibi = true;
     }
 
-    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);
-    }
+    hparams.rope_type = llama_rope_type(&model);
 }
 
 // TODO: This should probably be in llama.h
@@ -16252,7 +16225,7 @@ enum llama_vocab_type llama_vocab_type(const struct llama_model * model) {
     return model->vocab.type;
 }
 
-enum llama_rope_type llama_default_rope_type(const struct llama_model * model) {
+enum llama_rope_type llama_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
index 632136ca..16cece5d 100644
--- 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_default_rope_type   (const struct llama_model   * model);
+    LLAMA_API enum llama_rope_type    llama_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);