Skip to content

Bug: fix inputs to conv1d in mamba layer of plamo2 #14716

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jul 16, 2025
Merged

Bug: fix inputs to conv1d in mamba layer of plamo2 #14716

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion src/llama-model.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15763,6 +15763,7 @@ struct llm_build_plamo2 : public llm_graph_context_mamba {
cb(zx, "mamba_in_proj", il);
// {8192, 5, 1, 1} -> {8192, 1, 5, 1}
zx = ggml_permute(ctx0, zx, 0, 2, 1, 3);
zx = ggml_cont(ctx0, zx);
zx = ggml_reshape_4d(ctx0, zx, head_dim * 2, n_heads, n_seq_tokens, n_seqs);
cb(zx, "mamba_in_proj_out", il);

Expand All @@ -15780,7 +15781,6 @@ struct llm_build_plamo2 : public llm_graph_context_mamba {
// conv1d
{
// => {d_conv - 1 + n_seq_tokens, d_inner, n_seqs}
x = ggml_view_2d(ctx0, x, d_inner, n_seq_tokens * n_seqs, d_inner * x->nb[0], 0);
ggml_tensor * conv_x = ggml_concat(ctx0, conv, ggml_transpose(ctx0, x), 0);
cb(conv_x, "mamba_conv1d_input", il);

Expand Down
Loading