fix: update callback for ffn_moe_weighted and add callback for attn_out in deepseek2 model

Author: bluebread

src/llama-graph.cpp

@@ -1106,7 +1106,7 @@ ggml_tensor * llm_graph_context::build_moe_ffn(
 
     if (!weight_before_ffn) {
         experts = ggml_mul(ctx0, experts, weights);
-        cb(cur, "ffn_moe_weighted", il);
+        cb(experts, "ffn_moe_weighted", il);
     }
 
     ggml_tensor * cur_experts[LLAMA_MAX_EXPERTS] = { nullptr };

src/models/deepseek2.cpp

@@ -74,6 +74,7 @@ llm_build_deepseek2::llm_build_deepseek2(const llama_model & model, const llm_gr
             cur = build_attn(inp_attn,
                         model.layers[il].wo, NULL,
                         Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
+            cb(cur, "attn_out", il);
         }
         else {
             ggml_tensor * q = NULL;