Skip to content

Commit

Permalink
llama : switch KQ multiplication to F32 precision by default (#10015)
Browse files Browse the repository at this point in the history
ggml-ci
  • Loading branch information
ggerganov authored Oct 27, 2024
1 parent cc2983d commit 8841ce3
Showing 1 changed file with 4 additions and 11 deletions.
15 changes: 4 additions & 11 deletions src/llama.cpp
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

1 comment on commit 8841ce3

@ggerganov
Copy link
Owner Author

@ggerganov ggerganov commented on 8841ce3 Oct 27, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

https://github.com/ggerganov/llama.cpp/actions/runs/11543037894/job/32126951378#step:8:525

#10 15.50 ERROR: /requirements/requirements-convert_legacy_llama.txt recursively references itself in /requirements/requirements-convert_hf_to_gguf.txt and again in requirements.txt

Is this apip bug? This is not a recursive reference.

Please sign in to comment.