Various issues with basert pull

[bobosola]

basert pull basecompute/Qwen3.8-27B only allows downloading the pre-converted 4bit quant. There appears to be no (documented) way to download either the pre-converted 8 bit quant model or the MTP 4 bit quant model which are on your hf model page?

So I tried: basert pull Qwen/Qwen3.8-27B --target base-q8 and got this:

basert pull v0.2.3
  source:  Qwen/Qwen3.8-27B@main (HuggingFace, convert-on-pull)
  variant: default-q8
base-convert v0.2.3
  input:   /Volumes/990EVOPlus/Applications/baseRT/.src/hf/models--Qwen--Qwen3.8-27B/snapshots/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0
  output:  /Volumes/990EVOPlus/Applications/baseRT/Qwen/Qwen3.8-27B/default-q8/model.base
  profile: default-q8
Error: converting Qwen/Qwen3.8-27B

Caused by:
    0: opening shard "/Volumes/990EVOPlus/Applications/baseRT/.src/hf/models--Qwen--Qwen3.8-27B/snapshots/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/model-00001-of-00018.safetensors"
    1: opening "/Volumes/990EVOPlus/Applications/baseRT/.src/hf/models--Qwen--Qwen3.8-27B/snapshots/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/model-00001-of-00018.safetensors"
    2: No such file or directory (os error 2)

The converter seems to be looking for shard files named as model-00001-of-00018.safetensors but the actual downloaded model files are in the (55GB) blob folder with suffix-less GUID-type names thus:

[bobosola]

Just tried basert pull Qwen/Qwen3.8-27B --target base-q8 again with same result. It seems your downloader is failing to rename the model file names after the download completes?