Skip to content
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

[pull] master from ggerganov:master #150

Closed
wants to merge 2 commits into from
Closed

[pull] master from ggerganov:master #150

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8841ce3
llama : switch KQ multiplication to F32 precision by default (#10015)
ggerganov Oct 27, 2024
8125e6c
server : don't overfill the batch during infill (#10018)
ggerganov Oct 28, 2024
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
1 change: 1 addition & 0 deletions examples/server/server.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1880,6 +1880,7 @@ struct server_context {
if (slot.state == SLOT_STATE_STARTED) {
slot.t_start_process_prompt = ggml_time_us();
slot.t_start_generation = 0;

slot.n_past = 0;
slot.n_prompt_tokens = prompt_tokens.size();
slot.state = SLOT_STATE_PROCESSING_PROMPT;
Expand Down
6 changes: 4 additions & 2 deletions examples/server/utils.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -266,8 +266,10 @@ static llama_tokens format_infill(
}

// for now pick FIM context to fit in a batch (ratio prefix:suffix = 3:1, TODO: configurable?)
const int n_suffix_take = std::min<int>(tokens_suffix.size(), (n_batch/4));
const int n_prefix_take = std::min<int>(tokens_prefix.size(), 3*(n_batch/4) - 3);
const int n_prefix_take = std::min<int>(tokens_prefix.size(), 3*(n_batch/4));
const int n_suffix_take = std::min<int>(tokens_suffix.size(), std::max<int>(0, (n_batch/4) - (2 + tokens_prompt.size())));

SRV_DBG("n_prefix_take = %d, n_suffix_take = %d, total = %d\n", n_prefix_take, n_suffix_take, (n_prefix_take + n_suffix_take));

// fill the rest of the context with extra chunks
const int n_extra_take = std::min<int>(std::max<int>(0, n_ctx - (n_batch) - 2*n_predict), extra_tokens.size());
Expand Down
15 changes: 4 additions & 11 deletions src/llama.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9618,20 +9618,16 @@ static struct ggml_tensor * llm_build_kqv(
cur = ggml_flash_attn_ext(ctx, q, k, v, kq_mask, kq_scale, hparams.f_max_alibi_bias,
hparams.attn_soft_cap ? hparams.f_attn_logit_softcapping : 0.0f);

if (model.arch == LLM_ARCH_PHI2 || model.arch == LLM_ARCH_PHI3 || model.arch == LLM_ARCH_GPTNEOX || model.arch == LLM_ARCH_GEMMA2) {
ggml_flash_attn_ext_set_prec(cur, GGML_PREC_F32);
}
ggml_flash_attn_ext_set_prec(cur, GGML_PREC_F32);

cur = ggml_reshape_2d(ctx, cur, n_embd_head_v*n_head, n_tokens);
} else {
struct ggml_tensor * kq = ggml_mul_mat(ctx, k, q);
cb(kq, "kq", il);

if (model.arch == LLM_ARCH_PHI2 || model.arch == LLM_ARCH_PHI3 || model.arch == LLM_ARCH_GPTNEOX || model.arch == LLM_ARCH_QWEN2 || model.arch == LLM_ARCH_NEMOTRON || model.arch == LLM_ARCH_CHATGLM) {
// for this arch, we need to perform the KQ multiplication with F32 precision, otherwise we get NaNs
// ref: https://github.com/ggerganov/llama.cpp/pull/4490#issuecomment-1859055847
ggml_mul_mat_set_prec(kq, GGML_PREC_F32);
}
// note: this op tends to require high floating point range
// while for some models F16 is enough, for others it is not, so we default to F32 here
ggml_mul_mat_set_prec(kq, GGML_PREC_F32);

if (model.arch == LLM_ARCH_GROK) {
// need to do the following:
Expand All @@ -9640,9 +9636,6 @@ static struct ggml_tensor * llm_build_kqv(
// kq = 30 * tanh(kq / 30)
// before the softmax below

//try from phi2
//ggml_mul_mat_set_prec(kq, GGML_PREC_F32);

kq = ggml_tanh(ctx, ggml_scale(ctx, kq, 0.08838834764831845f/30.0f));
kq = ggml_scale(ctx, kq, 30);
}
Expand Down