vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

metal : fix nans in soft_max

Browse source
This commit is contained in:
Georgi Gerganov 2024-02-19 10:00:16 +02:00
parent 558d5d4b08
commit 76d9cbf60e
No known key found for this signature in database
GPG key ID: BF970631944C16B7
1 changed files with 4 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

8
ggml-metal.metal
View file

@ -392,7 +392,7 @@ kernel void kernel_soft_max(
float lmax = -INFINITY;
for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
lmax = MAX(lmax, psrc0[i00]*scale + (pmask ? pmask[i00] : 0.0f) + slope*ppos[i00]);
lmax = MAX(lmax, psrc0[i00]*scale + (pmask ? pmask[i00] : 0.0f) + (ppos ? slope*ppos[i00] : 0.0f));
}
// find the max value in the block
@ -417,7 +417,7 @@ kernel void kernel_soft_max(
// parallel sum
float lsum = 0.0f;
for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
const float exp_psrc0 = exp((psrc0[i00]*scale + (pmask ? pmask[i00] : 0.0f) + slope*ppos[i00]) - max_val);
const float exp_psrc0 = exp((psrc0[i00]*scale + (pmask ? pmask[i00] : 0.0f) + (ppos ? slope*ppos[i00] : 0.0f)) - max_val);
lsum += exp_psrc0;
pdst[i00] = exp_psrc0;
}
@ -495,7 +495,7 @@ kernel void kernel_soft_max_4(
float4 lmax4 = -INFINITY;
for (int i00 = tpitg; i00 < ne00/4; i00 += ntg) {
lmax4 = fmax(lmax4, psrc4[i00]*scale + (pmask ? pmask[i00] : 0.0f) + slope*ppos[i00]);
lmax4 = fmax(lmax4, psrc4[i00]*scale + (pmask ? pmask[i00] : 0.0f) + (ppos ? slope*ppos[i00] : 0.0f));
}
const float lmax = MAX(MAX(lmax4[0], lmax4[1]), MAX(lmax4[2], lmax4[3]));
@ -521,7 +521,7 @@ kernel void kernel_soft_max_4(
// parallel sum
float4 lsum4 = 0.0f;
for (int i00 = tpitg; i00 < ne00/4; i00 += ntg) {
const float4 exp_psrc4 = exp((psrc4[i00]*scale + (pmask ? pmask[i00] : 0.0f) + slope*ppos[i00]) - max_val);
const float4 exp_psrc4 = exp((psrc4[i00]*scale + (pmask ? pmask[i00] : 0.0f) + (ppos ? slope*ppos[i00] : 0.0f)) - max_val);
lsum4 += exp_psrc4;
pdst4[i00] = exp_psrc4;
}