keys ๐งช
@u1tra_instinct
#Bitcoin# ๐๐ . ๐งช๐งช๐งช (always testing and get into trouble).
๊ฐ์
January 2018
735 ํ๋ก์ ์ค 4.4K ํฌ
85toks/sec on Prose that is sickโฆ LFG
More DS4.1 TP4 improvements on 4x DGX Spark ๐
- Prose decode: 85 tok/s at c1 (was 69)
- Boot: ~3 min
- Prefill: ~4.8k tok/s at 32kโ128k, 4.3k at 262k
- KV cache: ~6.5M tokens, 1M context
- Same weights, same quality (qeval unchanged)
๋ ๋ณด๊ธฐ
