tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-26 19:55:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

context : remove logits_all flag (#13284)

Browse Source 
* context : remove logits_all flag

ggml-ci

* llama : remove logits_all flag + reorder llama_context_params

ggml-ci
This commit is contained in:
Georgi Gerganov
2025-05-08 14:26:50 +03:00
committed by GitHub
parent 70a6991edf
commit 51fb96b1ff
9 changed files with 13 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
common/arg.cpp
View File

@ -2097,13 +2097,6 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
params.cache_type_v = kv_cache_type_from_str(value);
}
).set_env("LLAMA_ARG_CACHE_TYPE_V"));
add_opt(common_arg(
{"--perplexity", "--all-logits"},
string_format("return logits for all tokens in the batch (default: %s)", params.logits_all ? "true" : "false"),
[](common_params & params) {
params.logits_all = true;
}
).set_examples({LLAMA_EXAMPLE_PERPLEXITY}));
add_opt(common_arg(
{"--hellaswag"},
"compute HellaSwag score over random tasks from datafile supplied with -f",

1
common/common.cpp
View File

@ -1096,7 +1096,6 @@ struct llama_context_params common_context_params_to_llama(const common_params &
cparams.n_threads = params.cpuparams.n_threads;
cparams.n_threads_batch = params.cpuparams_batch.n_threads == -1 ?
params.cpuparams.n_threads : params.cpuparams_batch.n_threads;
cparams.logits_all = params.logits_all;
cparams.embeddings = params.embedding;
cparams.rope_scaling_type = params.rope_scaling_type;
cparams.rope_freq_base = params.rope_freq_base;

1
common/common.h
View File

@ -324,7 +324,6 @@ struct common_params {
bool ctx_shift = true; // context shift on inifinite text generation
bool input_prefix_bos = false; // prefix BOS to user inputs, preceding input_prefix
bool logits_all = false; // return logits for all tokens in the batch
bool use_mmap = true; // use mmap for faster loads
bool use_mlock = false; // use mlock to keep model in memory
bool verbose_prompt = false; // print prompt tokens before generation