swift15: gate MNBT 3072 per Chief (w4 arena scales hard with MNBT; 32k step is absurd here)
This commit is contained in:
1 parent
f20da6a6ff
commit
79c2e64c2c
1 file changed
+1
-1
@@ -39,7 +39,7 @@ podman run --rm --name "$C" \
|
|||||||
--override-chat-template /models/tpl.jinja \
|
--override-chat-template /models/tpl.jinja \
|
||||||
--kld-ref /data/kld-ref --kld-out /data/kld-gate.json \
|
--kld-ref /data/kld-ref --kld-out /data/kld-gate.json \
|
||||||
--vram-kv-mib 2048 \
|
--vram-kv-mib 2048 \
|
||||||
--max-num-seqs 4 --max-num-batched-tokens 16384 \
|
--max-num-seqs 4 --max-num-batched-tokens 3072 \
|
||||||
2>&1 | tee "$LOG"
|
2>&1 | tee "$LOG"
|
||||||
rc=${PIPESTATUS[0]}
|
rc=${PIPESTATUS[0]}
|
||||||
[ "$rc" -eq 0 ] || { echo "KLD GATE BOOT FAILED rc=$rc; log: $LOG"; exit "$rc"; }
|
[ "$rc" -eq 0 ] || { echo "KLD GATE BOOT FAILED rc=$rc; log: $LOG"; exit "$rc"; }
|
||||||
|
|||||||
Reference in new issue
Block a user