Skip to content

Actions: drunkenbot-ai/engine

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
71 workflow runs
71 workflow runs

Filter by Workflow

⌥ + click/return to exclude

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
fix(hyperparameters): eliminate hardcoded 16GB VRAM assumptions and s… Engine code quality #71: Commit 4539dc9 pushed by drunkenbot-ai
10s main
main
10s
fix(recipe): standardize on scalable 11-pillar frontier base and remo… Engine code quality #70: Commit 4a56150 pushed by drunkenbot-ai
9s main
main
9s
feat(engine): add dataset scrubber, HF streaming presets, and curricu… Engine code quality #69: Commit 0369e1a pushed by drunkenbot-ai
10s main
main
10s
Implement Phase 3: Add Annealing step scheduling to training planning Engine code quality #68: Commit 61809e9 pushed by drunkenbot-ai
9s main
main
9s
Implement Phase 2: Add 1B frontier curriculum recipe and streaming pa… Engine code quality #67: Commit 9d372e0 pushed by drunkenbot-ai
10s main
main
10s
Implement Phase 1: Near-duplicate MinHash filtering, prefix boilerpla… Engine code quality #66: Commit 1e88271 pushed by drunkenbot-ai
15s main
main
15s
Optimize micro-batch size and gradient accumulation to target 85-90% … Engine code quality #65: Commit a19df99 pushed by drunkenbot-ai
9s main
main
9s
fix(attention): ensure contiguous query, key, value tensors before SD… Engine code quality #64: Commit ae2a3b5 pushed by drunkenbot-ai
11s main
main
11s
feat(engine): optimize VRAM planning calculations and export load_tok… Engine code quality #63: Commit 2afa435 pushed by drunkenbot-ai
9s main
main
9s
feat(optimizer): add dynamic training hyperparameter optimization for… Engine code quality #62: Commit 989963c pushed by drunkenbot-ai
10s main
main
10s
feat(recipe): synchronize 11-pillar recipe with v2.0-frontier-curated… Engine code quality #61: Commit b256d6c pushed by drunkenbot-ai
15s main
main
15s
feat(architecture): add support for untied word embeddings Engine code quality #60: Commit 9c92b06 pushed by drunkenbot-ai
14s main
main
14s
feat(dataset): remove legacy extraction heuristics and streamline ing… Engine code quality #59: Commit b0d9d54 pushed by drunkenbot-ai
13s main
main
13s
feat(recipe): fix locked token formula, add disk scanning, 11 pillars… Engine code quality #58: Commit 7847595 pushed by drunkenbot-ai
10s main
main
10s
feat(dataset): add dynamic dataset recipe engine, presets, and normal… Engine code quality #57: Commit e9a9ac3 pushed by drunkenbot-ai
11s main
main
11s
feat(thinking): add thinking fine-tune dataset stage and loader support Engine code quality #56: Commit c90f3a7 pushed by drunkenbot-ai
12s main
main
12s
fix(lora): support top-level attention and mlp module target names Engine code quality #55: Commit 0fcc861 pushed by drunkenbot-ai
12s main
main
12s
fix(engine): remove dangerous whole-disk recursive rglob search from … Engine code quality #54: Commit 813c3df pushed by drunkenbot-ai
19s main
main
19s
optmize core Engine code quality #53: Commit f284b2c pushed by drunkenbot-ai
10s main
main
10s
fix licensing, add docs Engine code quality #52: Commit ff541da pushed by drunkenbot-ai
10s main
main
10s
optimize instruction and conversation fine tune Engine code quality #51: Commit 4c46053 pushed by drunkenbot-ai
14s main
main
14s
fix dimentions warning Engine code quality #50: Commit dffd8d4 pushed by drunkenbot-ai
13s main
main
13s
fix pretraining issues Engine code quality #49: Commit df3c5b3 pushed by drunkenbot-ai
11s main
main
11s
fix triton related errors Engine code quality #48: Commit 2f45541 pushed by drunkenbot-ai
13s main
main
13s
implement triton Engine code quality #47: Commit 26c0e1a pushed by drunkenbot-ai
11s main
main
11s