falcon-h1: remove unused ssm_in_b and bad merge

Author: younesbelkada

src/llama-model.cpp

@@ -4560,7 +4560,6 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
                         /*SSM LAYERS*/
                         // ssm in
                         layer.ssm_in = create_tensor(tn(LLM_TENSOR_SSM_IN, "weight", i), {hidden_size, ssm_projection_size}, 0);
-                        layer.ssm_in_b = create_tensor(tn(LLM_TENSOR_SSM_IN, "bias", i), {n_embd, ssm_projection_size}, TENSOR_NOT_REQUIRED);
                         // ssm 1d conv
                         layer.ssm_conv1d = create_tensor(tn(LLM_TENSOR_SSM_CONV1D, "weight", i), {ssm_conv_kernel_size, ssm_conv_dim}, 0);
                         layer.ssm_conv1d_b = create_tensor(tn(LLM_TENSOR_SSM_CONV1D, "bias", i), {ssm_conv_dim}, TENSOR_NOT_REQUIRED);
@@ -4613,6 +4612,7 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
 
                     for (int i = 0; i < n_layer; ++i) {
                         auto & layer = layers[i];
+
                         layer.attn_norm = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, 0);
 
                         layer.wq = create_tensor(tn(LLM_TENSOR_ATTN_Q,   "weight", i), {n_embd, n_embd_head_k * n_head}, 0);

src/llama-model.h

@@ -225,9 +225,6 @@ struct llama_layer {
     struct ggml_tensor * ffn_down_enc = nullptr;
     struct ggml_tensor * ffn_up_enc   = nullptr;
 
-    // falcon_h1
-    struct ggml_tensor * ssm_in_b   = nullptr;
-
     // ff MoE
     struct ggml_tensor * ffn_gate_inp  = nullptr;
     struct ggml_tensor * ffn_gate_exps = nullptr;