dank/serverAppConfigs
1
0 Fork 0
Code Issues 4 Pull Requests Actions Packages Projects Releases Wiki Activity

swift15: gate shape MNBT 16384 (w4 arena at 32768 costs 33 GiB), kv 2048, headroom 512

Browse Source
This commit is contained in:
Frank committed 2026-10-08 03:16:45 -07:00
1 parent 058ea6cca8
commit f20da6a6ff
1 file changed
+3 -3
Download Patch File Download Diff File Expand all files Collapse all files
scripts/kld-gate-swift15.sh
+3 -3
View File
@@ -32,14 +32,14 @@ podman run --rm --name "$C" \
-v "$KLD_REF_DIR:/data/kld-ref:ro" \ -v "$KLD_REF_DIR:/data/kld-ref:ro" \
"stilldeadcode/radiance@$PIN" \ "stilldeadcode/radiance@$PIN" \
--model /models/rad/qwen4exp-swift15-w4dn8-i8-hc8m.rad \ --model /models/rad/qwen4exp-swift15-w4dn8-i8-hc8m.rad \
--host-pool-mib 40960 --deterministic --gpu-headroom-mib 768 \ --host-pool-mib 40960 --deterministic --gpu-headroom-mib 512 \
--num-speculative-tokens 0 \ --num-speculative-tokens 0 \
--kv-cache-dtype fp8 --max-model-len 16384 \ --kv-cache-dtype fp8 --max-model-len 16384 \
--ngram-placement ram \ --ngram-placement ram \
--override-chat-template /models/tpl.jinja \ --override-chat-template /models/tpl.jinja \
--kld-ref /data/kld-ref --kld-out /data/kld-gate.json \ --kld-ref /data/kld-ref --kld-out /data/kld-gate.json \
--vram-kv-mib 3072 \ --vram-kv-mib 2048 \
--max-num-seqs 4 --max-num-batched-tokens 32768 \ --max-num-seqs 4 --max-num-batched-tokens 16384 \
2>&1 | tee "$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"; }