From adff470c8a2258f6a53b8b684645366d96a3d059 Mon Sep 17 00:00:00 2001 From: younesbelkada Date: Tue, 8 Jul 2025 11:19:38 +0400 Subject: [PATCH] more cleanups and fixed conversion --- convert_hf_to_gguf.py | 2 +- src/llama-arch.cpp | 1 - src/llama-arch.h | 1 - 3 files changed, 1 insertion(+), 3 deletions(-) diff --git a/convert_hf_to_gguf.py b/convert_hf_to_gguf.py index d6cd575483..27bdf1fd99 100755 --- a/convert_hf_to_gguf.py +++ b/convert_hf_to_gguf.py @@ -6655,7 +6655,7 @@ class FalconH1Model(Mamba2Model): ## Mamba mixer params ## self.gguf_writer.add_ssm_conv_kernel(self.find_hparam(["conv_kernel", "d_conv"])) self.gguf_writer.add_ssm_group_count(self.n_group) - self.gguf_writer.add_ssm_inner_size(self.d_inner) + self.gguf_writer.add_ssm_inner_size(self.find_hparam(["mamba_d_ssm"])) self.gguf_writer.add_ssm_head_dim(d_head := self.find_hparam(["d_head"])) self.gguf_writer.add_ssm_time_step_rank(self.find_hparam(["n_heads"])) diff --git a/src/llama-arch.cpp b/src/llama-arch.cpp index f5cc8edb30..2db2abc607 100644 --- a/src/llama-arch.cpp +++ b/src/llama-arch.cpp @@ -1736,7 +1736,6 @@ static const std::map LLM_TENSOR_INFOS = { {LLM_TENSOR_CLS, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_MUL_MAT}}, {LLM_TENSOR_CLS_OUT, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_MUL_MAT}}, {LLM_TENSOR_OUTPUT_NORM, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_MUL}}, - {LLM_TENSOR_FINAL_NORM, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_MUL}}, {LLM_TENSOR_DEC_OUTPUT_NORM, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_MUL}}, {LLM_TENSOR_ENC_OUTPUT_NORM, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_MUL}}, {LLM_TENSOR_ROPE_FREQS, {LLM_TENSOR_LAYER_REPEATING, GGML_OP_ROPE}}, diff --git a/src/llama-arch.h b/src/llama-arch.h index f84cd217a0..2bcd2c3cde 100644 --- a/src/llama-arch.h +++ b/src/llama-arch.h @@ -390,7 +390,6 @@ enum llm_tensor { LLM_TENSOR_POS_NET_ATTN_K, LLM_TENSOR_POS_NET_ATTN_V, LLM_TENSOR_POS_NET_ATTN_OUT, - LLM_TENSOR_FINAL_NORM, }; enum llm_tensor_layer {