feat: initial support for llama.cpp

README.md

@@ -106,6 +106,7 @@ Typically finetunes of the base models below are supported as well.
 - [x] [ChatGLM3-6b](https://huggingface.co/THUDM/chatglm3-6b) + [ChatGLM4-9b](https://huggingface.co/THUDM/glm-4-9b)
 - [x] [SmolLM](https://huggingface.co/collections/HuggingFaceTB/smollm-6695016cad7167254ce15966)
 - [x] [EXAONE-3.0-7.8B-Instruct](https://huggingface.co/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct)
+- [x] [FalconMamba Models](https://huggingface.co/collections/tiiuae/falconmamba-7b-66b9a580324dd1598b0f6d4a)
 
 (instructions for supporting more models: [HOWTO-add-model.md](./docs/development/HOWTO-add-model.md))
 

convert_hf_to_gguf.py

@@ -2756,6 +2756,7 @@ class MambaModel(Model):
         self.gguf_writer.add_ssm_state_size(d_state)
         self.gguf_writer.add_ssm_time_step_rank(dt_rank)
         self.gguf_writer.add_layer_norm_rms_eps(rms_norm_eps)
+        self.gguf_writer.add_mamba_b_dt_rms(False) # For classic Mamba we don't apply rms norm on B / DT layers
         self.gguf_writer.add_file_type(self.ftype)
 
     _tok_embd = None
@@ -3854,6 +3855,39 @@ class ExaoneModel(Model):
                 self.gguf_writer.add_tensor(self.format_tensor_name(gguf.MODEL_TENSOR.ROPE_FREQS), np.array(rope_factors, dtype=np.float32))
 
         super().prepare_tensors()
+        
+
+@Model.register("FalconMambaForCausalLM")
+class FalconMambaModel(MambaModel):
+    model_arch = gguf.MODEL_ARCH.MAMBA
+
+    def set_gguf_parameters(self):
+        d_model = self.find_hparam(["hidden_size",       "d_model"])
+        d_conv  = self.find_hparam(["conv_kernel",       "d_conv"],  optional=True) or 4
+        d_inner = self.find_hparam(["intermediate_size", "d_inner"], optional=True) or 2 * d_model
+        d_state = self.find_hparam(["state_size",        "d_state"], optional=True) or 16
+        # ceiling division
+        # ref: https://stackoverflow.com/a/17511341/22827863
+        # ref: https://github.com/state-spaces/mamba/blob/ce59daea3a090d011d6476c6e5b97f6d58ddad8b/mamba_ssm/modules/mamba_simple.py#L58
+        dt_rank      = self.find_hparam(["time_step_rank",     "dt_rank"],      optional=True) or -(d_model // -16)
+        rms_norm_eps = self.find_hparam(["layer_norm_epsilon", "rms_norm_eps"], optional=True) or 1e-5
+
+        # Fail early for models which don't have a block expansion factor of 2
+        assert d_inner == 2 * d_model
+
+        self.gguf_writer.add_context_length(2**20) # arbitrary value; for those who use the default
+        self.gguf_writer.add_embedding_length(d_model)
+        self.gguf_writer.add_feed_forward_length(0) # unused, but seemingly required when loading
+        self.gguf_writer.add_head_count(0) # unused, but seemingly required when loading
+        self.gguf_writer.add_block_count(self.hparams["num_hidden_layers"])
+        self.gguf_writer.add_ssm_conv_kernel(d_conv)
+        self.gguf_writer.add_mamba_b_dt_rms(True) # For FalconMamba we do apply rms norm on B / DT layers
+        self.gguf_writer.add_ssm_inner_size(d_inner)
+        self.gguf_writer.add_ssm_state_size(d_state)
+        self.gguf_writer.add_ssm_time_step_rank(dt_rank)
+        self.gguf_writer.add_layer_norm_rms_eps(rms_norm_eps)
+        self.gguf_writer.add_file_type(self.ftype)
+
 
 ###### CONVERSION LOGIC ######
 

gguf-py/gguf/constants.py

@@ -130,6 +130,7 @@ class Keys:
         INNER_SIZE     = "{arch}.ssm.inner_size"
         STATE_SIZE     = "{arch}.ssm.state_size"
         TIME_STEP_RANK = "{arch}.ssm.time_step_rank"
+        B_DT_RMS = "{arch}.ssm.b_dt_rms"
 
     class Tokenizer:
         MODEL                = "tokenizer.ggml.model"
@@ -1372,6 +1373,7 @@ KEY_SSM_CONV_KERNEL    = Keys.SSM.CONV_KERNEL
 KEY_SSM_INNER_SIZE     = Keys.SSM.INNER_SIZE
 KEY_SSM_STATE_SIZE     = Keys.SSM.STATE_SIZE
 KEY_SSM_TIME_STEP_RANK = Keys.SSM.TIME_STEP_RANK
+KEY_SSM_B_DT_RMS = Keys.SSM.B_DT_RMS
 
 # tokenization
 KEY_TOKENIZER_MODEL      = Keys.Tokenizer.MODEL

gguf-py/gguf/gguf_writer.py

@@ -714,6 +714,9 @@ class GGUFWriter:
 
     def add_rope_scaling_finetuned(self, value: bool) -> None:
         self.add_bool(Keys.Rope.SCALING_FINETUNED.format(arch=self.arch), value)
+        
+    def add_mamba_b_dt_rms(self, value: bool) -> None:
+        self.add_bool(Keys.SSM.B_DT_RMS.format(arch=self.arch), value)
 
     def add_rope_scaling_yarn_log_mul(self, value: float) -> None:
         self.add_float32(Keys.Rope.SCALING_YARN_LOG_MUL.format(arch=self.arch), value)

src/llama.cpp

@@ -328,6 +328,7 @@ enum llm_kv {
     LLM_KV_SSM_CONV_KERNEL,
     LLM_KV_SSM_STATE_SIZE,
     LLM_KV_SSM_TIME_STEP_RANK,
+    LLM_KV_SSM_B_DT_RMS,
 
     LLM_KV_TOKENIZER_MODEL,
     LLM_KV_TOKENIZER_PRE,
@@ -426,6 +427,7 @@ static const std::map<llm_kv, const char *> LLM_KV_NAMES = {
     { LLM_KV_SSM_INNER_SIZE,                "%s.ssm.inner_size"     },
     { LLM_KV_SSM_STATE_SIZE,                "%s.ssm.state_size"     },
     { LLM_KV_SSM_TIME_STEP_RANK,            "%s.ssm.time_step_rank" },
+    { LLM_KV_SSM_B_DT_RMS,                  "%s.ssm.b_dt_rms" },
 
     { LLM_KV_TOKENIZER_MODEL,                "tokenizer.ggml.model"                    },
     { LLM_KV_TOKENIZER_PRE,                  "tokenizer.ggml.pre"                      },
@@ -2237,6 +2239,7 @@ struct llama_hparams {
     uint32_t ssm_d_inner = 0;
     uint32_t ssm_d_state = 0;
     uint32_t ssm_dt_rank = 0;
+    bool ssm_b_dt_rms = false;
 
     float f_clamp_kqv      = 0.0f;
     float f_max_alibi_bias = 0.0f;
@@ -5052,6 +5055,7 @@ static void llm_load_hparams(
                 ml.get_key(LLM_KV_SSM_INNER_SIZE,     hparams.ssm_d_inner);
                 ml.get_key(LLM_KV_SSM_STATE_SIZE,     hparams.ssm_d_state);
                 ml.get_key(LLM_KV_SSM_TIME_STEP_RANK, hparams.ssm_dt_rank);
+                ml.get_key(LLM_KV_SSM_B_DT_RMS, hparams.ssm_b_dt_rms);
 
                 ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
 
@@ -12161,6 +12165,10 @@ struct llm_build_context {
         GGML_ASSERT(2 * d_model == d_inner);
         const int64_t d_state = hparams.ssm_d_state;
         const int64_t dt_rank = hparams.ssm_dt_rank;
+        // Some variants of Mamba arch (e.g. FalconMamba do apply layer norm on B and Dt layers)
+        const bool ssm_b_dt_rms = hparams.ssm_b_dt_rms;
+        // Use the same RMS norm as the final layer norm
+        const float norm_rms_eps = hparams.f_norm_rms_eps;
 
         struct ggml_tensor * cur;
         struct ggml_tensor * inpL;
@@ -12241,6 +12249,13 @@ struct llm_build_context {
                 struct ggml_tensor * B  = ggml_view_2d(ctx0, x_db, d_state, n_tokens, x_db->nb[1], ggml_element_size(x_db)*dt_rank);
                 struct ggml_tensor * C  = ggml_view_2d(ctx0, x_db, d_state, n_tokens, x_db->nb[1], ggml_element_size(x_db)*(dt_rank+d_state));
 
+                // Some Mamba variants (e.g. FalconMamba) apply RMS norm in B, C & Dt layers
+                if (ssm_b_dt_rms) {
+                    dt = ggml_rms_norm(ctx0, dt, norm_rms_eps);
+                    B = ggml_rms_norm(ctx0, B, norm_rms_eps);
+                    C = ggml_rms_norm(ctx0, C, norm_rms_eps);
+                }
+
                 // {dt_rank, d_inner} * {dt_rank, n_tokens} => {d_inner, n_tokens}
                 dt = llm_build_lora_mm(lctx, ctx0, model.layers[il].ssm_dt, dt);
                 dt = ggml_add(ctx0, dt, model.layers[il].ssm_dt_b);