v0.13.13

fix: wrong output layer offload at zero ts input

Signed-off-by: thxCode <[email protected]>

v0.13.12

refactor: estimate

Signed-off-by: thxCode <[email protected]>

v0.13.11

docs: readme

Signed-off-by: thxCode <[email protected]>

v0.13.10

fix: calculation error of projector zero offloading

Signed-off-by: thxCode <[email protected]>

v0.13.9

feat: support offloading sd to multi devs

Signed-off-by: thxCode <[email protected]>

v0.13.8

refactor: support qk_m distributable

Signed-off-by: thxCode <[email protected]>

v0.13.7

refactor: estimate

Signed-off-by: thxCode <[email protected]>

v0.13.6

refactor: sd arch

Signed-off-by: thxCode <[email protected]>

v0.13.5

refactor: embedding usage estimate

Signed-off-by: thxCode <[email protected]>

v0.13.4

refactor: new ggml type

Signed-off-by: thxCode <[email protected]>