nodes/city96-ComfyUI-GGUF
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Keep img_emb in F32 for wan i2v

Browse Source
This commit is contained in:
City
2025-02-27 05:58:01 +01:00
parent 32a8eab8f9
commit 929a9e5f0d
1 changed files with 5 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/lcpp.patch
+5 -4
View File
@@ -39,7 +39,7 @@ index b16c462f..6d1568f1 100644
const int idx = gguf_find_tensor(ctx, name);
if (idx < 0) {
diff --git a/src/llama.cpp b/src/llama.cpp
index 24e1f1f0..f90738cf 100644
index 24e1f1f0..d3ca4756 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -205,6 +205,14 @@ enum llm_arch {
@@ -296,7 +296,7 @@ index 24e1f1f0..f90738cf 100644
}
// Set split info if needed
@@ -18647,6 +18850,82 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
@@ -18647,6 +18850,83 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
// do not quantize relative position bias (T5)
quantize &= name.find("attn_rel_b.weight") == std::string::npos;
@@ -369,6 +369,7 @@ index 24e1f1f0..f90738cf 100644
+ quantize &= name.find("text_embedding.") == std::string::npos;
+ quantize &= name.find("time_projection.") == std::string::npos;
+ quantize &= name.find("time_embedding.") == std::string::npos;
+ quantize &= name.find("img_emb.") == std::string::npos;
+ quantize &= name.find("head.") == std::string::npos;
+ }
+ // ignore 3D/4D tensors for image models as the code was never meant to handle these
@@ -379,7 +380,7 @@ index 24e1f1f0..f90738cf 100644
enum ggml_type new_type;
void * new_data;
size_t new_size;
@@ -18655,6 +18934,9 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
@@ -18655,6 +18935,9 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
new_type = default_type;
// get more optimal quantization type based on the tensor shape, layer, etc.
@@ -389,7 +390,7 @@ index 24e1f1f0..f90738cf 100644
if (!params->pure && ggml_is_quantized(default_type)) {
new_type = llama_tensor_get_type(qs, new_type, tensor, ftype);
}
@@ -18664,6 +18946,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
@@ -18664,6 +18947,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
if (params->output_tensor_type < GGML_TYPE_COUNT && strcmp(tensor->name, "output.weight") == 0) {
new_type = params->output_tensor_type;
}