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

ci : integrate with ggml-org/ci (#2250)

Browse source 
* ci : run ctest

ggml-ci

* ci : add open llama 3B-v2 tests

ggml-ci

* ci : disable wget progress output

ggml-ci

* ci : add open llama 3B-v2 tg tests for q4 and q5 quantizations

ggml-ci

* tests : try to fix tail free sampling test

ggml-ci

* ci : add K-quants

ggml-ci

* ci : add short perplexity tests

ggml-ci

* ci : add README.md

* ppl : add --chunks argument to limit max number of chunks

ggml-ci

* ci : update README
This commit is contained in:
Georgi Gerganov 2023-07-18 14:24:43 +03:00 • committed by GitHub
parent 6cbf9dfb32
commit d01bccde9f
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
8 changed files with 312 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

15
llama.cpp
View file

@ -2024,9 +2024,18 @@ void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array *
}
// Normalize the second derivatives
float second_derivatives_sum = std::accumulate(second_derivatives.begin(), second_derivatives.end(), 0.0f);
for (float & value : second_derivatives) {
value /= second_derivatives_sum;
{
const float second_derivatives_sum = std::accumulate(second_derivatives.begin(), second_derivatives.end(), 0.0f);
if (second_derivatives_sum > 1e-6f) {
for (float & value : second_derivatives) {
value /= second_derivatives_sum;
}
} else {
for (float & value : second_derivatives) {
value = 1.0f / second_derivatives.size();
}
}
}
float cum_sum = 0.0f;