5 commits

Author	SHA1	Message	Date
Azure-Tang	203b853c75	rm KMoEGateDeepSeekV3, fall back to KMoEGate	2025-04-01 07:13:05 +00:00
Atream	a889288fc1	use compile for gate, slight performance improvement	2025-03-14 12:43:28 +00:00
Atream	5ec33d046d	optimize gguf dequant, save mem, support Q2_K ... use marlin for lm_head, lm_head only calc last token for prefill extend context window to 19K for DeepSeek-V3/R1 within 24GB VRAM	2025-02-22 06:13:01 +00:00
Atream	c189d55bd1	toy support for experts on GPU, no CUDA Graph	2025-02-15 15:16:00 +00:00
Azure	b7653b9c4f	add V3/R1 8 gpu yaml example	2025-02-14 02:56:13 +00:00