Chat now sends stream=true and renders SSE tokens incrementally with live
tok/s status, a stop button (AbortController), and a blinking cursor; because
streamed requests emit tracker 'proxy progress' events, the Call wall now
shows in-flight requests with live TPS too. Chat colors route through :root
tokens instead of hardcoded hex values.
ADR-0020 documents the changes and the mixed-topology routing flaw: a partial
GPU head (0-21) + full CPU node (0-39) gets downstream start_layer=0 instead
of 22, corrupting activations into 1-token generations that were billed and
polluted throughput stats. Fix steps recorded, not yet implemented.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Bump the node package's transformers floor to 5.12 (older versions lack
composite Qwen3_5MoeConfig handling and fail with missing vocab_size), and
explain in QUICKSTART/INSTALL_WINDOWS that the flash-linear-attention /
causal-conv1d fast-path warning is a harmless CPU fallback.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Qwen3.5/3.6-MoE checkpoints ship vision (model.visual.*) and multi-token-
prediction (mtp.*) weights; the partial shard loader assigned them into the
text-only Qwen3_5MoeForCausalLM and crashed with AttributeError 'mtp'.
Filter selected tensors against the built model's state_dict keys, matching
transformers' _keys_to_ignore_on_load_unexpected behavior.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Start background re-registration when the tracker is unreachable at startup so nodes do not stay permanently unregistered.
Co-authored-by: Cursor <cursoragent@cursor.com>
Unify --model and --model-id so catalog names use the tracker path, and allow --shard-start/--shard-end with --model instead of requiring --model-id.
Co-authored-by: Cursor <cursoragent@cursor.com>
Resolve conflicts in dashboard.html (Call wall + live TPS/queue from remote) and server.py (proxy progress logging, request id forwarding, current_requests on node entries).
Co-authored-by: Cursor <cursoragent@cursor.com>
