Skip to content

Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦

C

Merge pull request #1792 into integ/release-batch-5

JustVugga33d495aabbf
diff --cc c/glm53.c
index ff005627,77795a03..4aa767bd
--- a/c/glm53.c
+++ b/c/glm53.c
@@@ -4179,9 -3927,9 +4220,10 @@@ static int serve_one(GModel *m, Tok *to
rss_gb(), prompt_tokens, limited);
#ifdef COLI_VULKAN
glm53_vk_report(m, "turn");
+ g53c_report(); /* the dense chain's line, when it ran */
#endif
free(sequence);
+ free(keys);
return input_eof ? -1 : 0;
}