Few Lost Sheep...

#1222
by DavidAU - opened
  • Qwen2.5-Microsoft-NextCoder-Soar-Instruct-FUSED-CODER-Fast-11B we already did
  • Qwen3-Instruct-6B-Brainstorm20x-128k-ctx and Qwen3-Instruct-F16-6B-Brainstorm20x-128k-ctx failed with submitting with no architectures entry (, or } expected while parsing object/hash, at character offset 608 (before ""rope_theta": 100000...") at /llmjob/share/bin/llmjob line 1639. probably due to corrupted config.yaml.
  • All others I queued sucessfully.

You can check the status under https://hf.tst.eu/status.html

Oh god, config.yaml is a thing now? We are doomed, I say!

@nicoboss @mradermacher
My bad ; fixed the config.json file for each - should be good to go.

Sorry about that!
Thanks so much.

.. Yaml... that is so 2024 ish.

oh - one more just uploaded...

DavidAU/Qwen3-42B-A3B-2507-TOTAL-RECALL-v2-Medium-MASTER-CODER

Little late to the party.

  • Qwen2.5-Microsoft-NextCoder-Soar-Instruct-FUSED-CODER-Fast-11B actually failed because of a lack of tokenizer.model so while it already was queued it did not get processed sucessfully.
  • Qwen3-Instruct-6B-Brainstorm20x-128k-ctx and Qwen3-Instruct-F16-6B-Brainstorm20x-128k-ctx I now sucessfully queued.
  • Qwen3-42B-A3B-2507-TOTAL-RECALL-v2-Medium-MASTER-CODER is now queued as well.

Oh god, config.yaml is a thing now? We are doomed, I say!

Luckely I was mistaken and just missremembered the file type fo the config.json as simular data is inside the README.md metadata as YAML. Am I the only one that likes the simplicity of YAML? There now even is YAML Path simular to JSONPath and XPath. In totaly YAML fashion the YAML Path Standard got defined by some random person and now is the de-facto standard everyone uses: https://github.com/wwkimball/yamlpath/wiki/Segments-of-a-YAML-Path

Thank you again.

RE: Soar ; odd - there is no tokenizer.model for this one, nor a lot of other ones (??).
Tried re-uploading json files - no new commits.

Will try re-upload of the safetensor files in am local time ; in case one of these glitched.

Am I the only one that likes the simplicity of YAML?

Must be - it's got one of the most complicated specs that exists on this planet, and still can't decide whether no is a string or a boolean. There is nothing simple when a format isn't even interoperable between different parsers, all spec-compliant...

There would objectively be a lot less grief in the world if YAML didn't exist.

Sign up or log in to comment