=== START 12:09:51
/dev/nvme0n1p2  916G  750G  120G   87% /
ds4: Linux cuda backend set oom_score_adj=1000
ds4: CUDA backend initialized on NVIDIA GB10 (sm_121) dev=0
ds4: SSD streaming mixed-precision model: 10/43 routed layers off the slab size class will bypass the expert cache and read experts via mapped model views
ds4: SSD streaming initial cuda model map restricted to token embedding (1 spans, 0.99 GiB tensor span)
ds4: CUDA host registration skipped: operation not supported
ds4: CUDA preparing model tensor mappings
ds4: CUDA loading model tensors into device cache
ds4: CUDA startup model preparation covered 0.99 GiB of tensor spans in 0.445s
ds4: CUDA directly mapped 0.87 GiB auxiliary model
ds4: cuda backend initialized for graph diagnostics
ds4: memory: KV 0.39 GiB (raw 0.34 + compressed 0.05) + buffers 0.03 GiB + resident model 3.36 GiB = 3.78 GiB planned
ds4: memory detail: ctx=4096 prefill_cap=4096 raw_kv_rows=4096 compressed_kv_rows=1026 backend=cuda
ds4: memory: KV 0.39 GiB (raw 0.34 + compressed 0.05) + buffers 0.03 GiB = 0.42 GiB context
ds4: memory detail: ctx=4096 prefill_cap=4096 raw_kv_rows=4096 compressed_kv_rows=1026 backend=cuda
ds4: CUDA loading model tensors 7.33 GiB cached
The user wants me to reply with exactly "LOADGATE OK". That's the only instruction. So I'll just output that exact phrase.
LOADGATE OK
ds4: prefill: 0.91 t/s, generation: 1.11 t/s
=== RC=0
=== END 12:10:44