|
first LoRA that actually helped
|
10-19-2026, 02:49 AM
Quote:Maybe, but I had the opposite result. I've seen people blame VRAM headroom when the real issue was much more mundane. Maybe, but I had the opposite result. I get why that seems right, but on GB200 the bottleneck felt elsewhere. Could be a workload difference, but I wouldn't generalize too fast from one setup.
throughput matters, but tail latency matters too
|
|
« Next Oldest | Next Newest »
|
| Messages In This Thread |
|
first LoRA that actually helped - by benchplease - 10-16-2026, 08:55 PM
Re: first LoRA that actually helped - by benchplease - 10-18-2026, 09:14 PM
Re: first LoRA that actually helped - by batchsize - 10-18-2026, 09:24 PM
Re: first LoRA that actually helped - by localtinker - 10-18-2026, 09:39 PM
Re: first LoRA that actually helped - by cacheline - 10-19-2026, 12:00 AM
Re: first LoRA that actually helped - by kernelghost - 10-19-2026, 12:01 AM
Re: first LoRA that actually helped - by cacheline - 10-19-2026, 01:26 AM
Re: first LoRA that actually helped - by batchsize - 10-19-2026, 02:49 AM
Re: first LoRA that actually helped - by benchplease - 10-19-2026, 05:23 AM
|
