Skip to content

Feature/403 error loading#416

Merged
michalharakal merged 3 commits intodevelopfrom
feature/403-error-loading
Mar 4, 2026
Merged

Feature/403 error loading#416
michalharakal merged 3 commits intodevelopfrom
feature/403-error-loading

Conversation

@michalharakal
Copy link
Contributor

No description provided.

michalharakal and others added 3 commits March 4, 2026 09:46
Refactor the monolithic model loading architecture by extracting
model-specific runtime, weight loading, and configuration code from
skainet-apps and skainet-io-gguf into dedicated per-model modules:

- skainet-model-llama: LlamaRuntime, weight loaders, config parser
- skainet-model-gemma: Gemma3n runtime, attention, KV cache, encoders
- skainet-model-bert: BertRuntime, tokenizer, weight loader
- skainet-model-qwen: Qwen config parser, weight source

Shared abstractions (ModelArchitecture, QuantPolicy) moved to
skainet-io-core. All app modules updated to depend on the new
model modules with updated package imports.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace directory-scan-only approach with an explicit smoke-models.json
config that supports kllama, kgemma, and kbert runners with per-model
overrides. Legacy folder-scan and CLI usage is preserved as a fallback.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@michalharakal michalharakal merged commit bdb3eaf into develop Mar 4, 2026
4 checks passed
@michalharakal michalharakal deleted the feature/403-error-loading branch March 6, 2026 19:47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant