Skip to content

fix(decode): reserve >= dflash_block_size outputs for DFlash draft contexts (#108) #114

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

Closed
marksverdhei wants to merge 1 commit into from
Closed

fix(decode): reserve >= dflash_block_size outputs for DFlash draft contexts (#108) #114

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
9 changes: 9 additions & 0 deletions src/llama-context.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -208,6 +208,15 @@ llama_context::llama_context(

cparams.n_outputs_max = params.n_outputs_max == 0 || llama_model_has_encoder(&model) ? cparams.n_batch : params.n_outputs_max;

// DFlash decodes an entire diffusion block per step with every token flagged
// for output (block_size tokens), so the output buffer must hold at least the
// block size. The server sizes draft contexts to n_parallel (server-context.cpp),
// which is < dflash_block_size and trips GGML_ASSERT(n_outputs_max <= cparams.n_outputs_max)
// in output_reserve() on the first draft(). See heiervang-technologies/ht-llama.cpp#108.
if (model.arch == LLM_ARCH_DFLASH) {
cparams.n_outputs_max = std::max(cparams.n_outputs_max, model.hparams.dflash_block_size);
}

cparams.op_offload = params.op_offload;
cparams.kv_unified = params.kv_unified;

Expand Down
Loading