popov/neuron-tai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

dash QOL

Browse Source
This commit is contained in:
Dobromir Popov
2026-07-07 17:37:38 +03:00
parent 640ef78711
commit e81d989f39
12 changed files with 1392 additions and 358 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
docs/issues/44-tracker-shard-source-partial-download.md
View File

@@ -75,7 +75,7 @@ What exists already (build on it, don't duplicate):
- [x] Node downloader keeps exact-shard peers first, then races tracker model
sources against a HuggingFace `snapshot_download(..., allow_patterns=...)`
subset download, using the first successful source.
- [ ] When no tracker model source is available at all, the HuggingFace
- [x] When no tracker model source is available at all, the HuggingFace
fallback still computes `allow_patterns` from the repo's own
`model.safetensors.index.json` (fetched directly, not via the tracker) —
it never silently downloads the full model just because the tracker has
@@ -95,7 +95,9 @@ What exists already (build on it, don't duplicate):
- 2026-07-06: Added the tracker/node download path. For immediate Qwen3.6-35B
LAN testing, real PyTorch nodes fetch the full snapshot from the tracker via
`full_url` and race HuggingFace as fallback. Remaining hard half is true
partial model materialization: the backend can prefer a downloaded local
model directory, but Transformers still needs a `meta`-device load path that
materializes only assigned layers.
`full_url`; HuggingFace remains fallback-only, and when it is used the node
computes `allow_patterns` from the repo's remote SafeTensors index so it
stays layer-filtered even without tracker-cached files. Remaining hard half
is true partial model materialization: the backend can prefer a downloaded
local model directory, but Transformers still needs a `meta`-device load
path that materializes only assigned layers.