Dobromir Popov 471893c9d5 Skip multimodal/MTP checkpoint tensors absent from the text-only causal LM ...

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>

2026-07-07 19:16:19 +02:00

..

__init__.py

feat: scaffold meshnet monorepo

2026-06-29 00:28:29 +03:00

cli.py

Fix meshnet-node model and shard flag parsing.

2026-07-07 17:54:30 +02:00

config.py

new tasks, model pricing, auto quantisation, etc...

2026-07-06 17:11:53 +03:00

dashboard.py

node stats and benchmark, dynamic realocation working

2026-07-01 10:02:17 +03:00

downloader.py

dash QOL

2026-07-07 17:37:38 +03:00

hardware.py

Record CUDA benchmark diagnostics

2026-07-01 10:57:44 +02:00

model_backend.py

Skip multimodal/MTP checkpoint tensors absent from the text-only causal LM

2026-07-07 19:16:19 +02:00

model_catalog.py

fix model registration and anouncement. added console panel

2026-07-07 14:24:37 +02:00

model_metadata.json

feat(alpha): complete hardening backlog

2026-07-05 21:47:23 +03:00

relay_bridge.py

Log node request processing so operators can see live activity in the console.

2026-07-07 18:12:57 +02:00

safetensors_selection.py

dash QOL

2026-07-07 17:37:38 +03:00

server.py

feat: tracker-as-first-layer-node inference entry point (US-014)

2026-06-29 16:59:32 +03:00

startup.py

try fix model loading quen3.6-35b

2026-07-07 18:36:29 +02:00

torch_server.py

set max tokens to 5k

2026-07-07 18:21:13 +02:00

wallet.py

feat: add node startup flow

2026-06-29 01:30:12 +03:00

wizard.py

inference working

2026-06-29 23:54:35 +03:00