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

llama : remove token functions with context args in favor of model (#3720)

Browse source 
* added `llama_model_token_*` variants to all the `llama_token_*` functions.

* added `LLAMA_API`

* formatting

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* removed old `llama_token` functions

* changed 3 more functions to take in model

- `llama_token_get_text`
- `llama_token_get_score`
- `llama_token_get_type`

* added back docs

* fixed main.cpp

* changed token functions to use new model variants

* changed token functions to use new model variants

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Marcus Dunn 2023-10-23 12:40:03 -07:00 • committed by GitHub
parent 6336701c93
commit 5be6c803fa
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
16 changed files with 81 additions and 79 deletions
Download patch file Download diff file Expand all files Collapse all files

4
common/sampling.cpp
View file

@ -147,7 +147,7 @@ llama_token llama_sampling_sample(
// apply penalties
if (!prev.empty()) {
const float nl_logit = logits[llama_token_nl(ctx_main)];
const float nl_logit = logits[llama_token_nl(llama_get_model(ctx_main))];
llama_sample_repetition_penalties(ctx_main, &cur_p,
prev.data() + prev.size() - penalty_last_n,
@ -155,7 +155,7 @@ llama_token llama_sampling_sample(
if (!penalize_nl) {
for (size_t idx = 0; idx < cur_p.size; idx++) {
if (cur_p.data[idx].id == llama_token_nl(ctx_main)) {
if (cur_p.data[idx].id == llama_token_nl(llama_get_model(ctx_main))) {
cur_p.data[idx].logit = nl_logit;
break;
}