llama : fix defrag logic (#11707)

* llama : fix defrag logic

ggml-ci

* cont : better logic

ggml-ci

* cont : clamp fragmentation to 0.0

ggml-ci
This commit is contained in:
Georgi Gerganov 2025-02-07 16:05:34 +02:00 committed by GitHub
parent 2d219b389e
commit ed926d8833
No known key found for this signature in database
GPG key ID: B5690EEEBB952194

View file

@ -8801,12 +8801,14 @@ static int llama_decode_impl(
//llama_synchronize(&lctx);
// decide if we need to defrag the kv cache
if (cparams.causal_attn && cparams.defrag_thold >= 0.0f) {
const float fragmentation = kv_self.n >= 128 ? 1.0f - float(kv_self.used)/float(kv_self.n) : 0.0f;
if (cparams.causal_attn && cparams.defrag_thold > 0.0f) {
// - do not defrag small contexts (i.e. < 2048 tokens)
// - count the padding towards the number of used tokens
const float fragmentation = kv_self.n >= 2048 ? std::max(0.0f, 1.0f - float(kv_self.used + llama_kv_cache_get_padding(cparams))/float(kv_self.n)) : 0.0f;
// queue defragmentation for next llama_kv_cache_update
if (fragmentation > cparams.defrag_thold) {
//LLAMA_LOG_INFO("fragmentation: %.2f\n", fragmentation);
LLAMA_LOG_DEBUG("%s: fragmentation: %.2f - requesting defrag\n", __func__, fragmentation);
llama_kv_cache_defrag(kv_self);
}