Hacker News
new
past
comments
ask
show
jobs
points
by
erwan577
2 hours ago
|
comments
by
verdverm
1 hours ago
|
[-]
quanting kv cache hurts attention / recall, and long-form tasks by proxy. Model families and sizes have different tolerances to quant ting different parts of the model, same for intended tasks.
reply