Joseph Magly
9746c21b63
test: enforce Wave A quality lock
2026-08-14 19:47:46 -04:00
Joseph Magly
c1a1275291
ci: add conditional environment test gates
2026-08-14 14:29:14 -04:00
Joseph Magly
951700a285
test: add quality-depth gates and mature CPU coverage
2026-08-14 14:08:23 -04:00
Joseph Magly
8bbb5f2926
test: add offline integration baseline
2026-08-14 13:15:32 -04:00
Joseph Magly
6add02edd0
test: enforce model and CLI boundary contracts
2026-08-14 12:51:09 -04:00
Joseph Magly
2e703e19e2
docs: separate Hugging Face Space metadata
2026-08-14 10:37:08 -04:00
faber and Claude Opus 4.6
04b8ec60cb
Add ASPA framework, AutoObliterator, Watchtower, expanded eval corpus
...
New core modules:
- auto_obliterate.py: Automated multi-iteration obliteration pipeline
- watchtower.py: HF Hub model discovery and tracking
- ui_watchtower.py: Gradio tabs for Watchtower (ready for app.py wiring)
- hard_negative.py: Residue mining from refusal audits
- model_profile.py: Parameter profiling from safetensors/config
- bestiary_sync.py: Sync models from PlinyOS BESTIARY registry
- models_client.py: Lightweight HF model list client
Framework enhancements:
- abliterate.py: ASPA source-tethering, step gradient blending, hard-negative residue support
- cli.py: self-improve command, model profiling, hard-negative flags
- prompts.py: Expanded 842-prompt refusal eval corpus across 10 categories
- __init__.py: New exports (Watchtower, AutoObliterator)
Reference implementations (14 scripts):
- ASPA sweep, gradient search, coherence eval, MMLU benchmarks
- Pareto controller, refusal sniper, stock comparisons
Documentation:
- README: Research framing, responsible use section, comprehensive disclaimer
- docs/beyond_sota_roadmap.md, docs/recursive_self_improvement.md
Tests: 4 new test files (354 lines)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com >
2026-06-09 03:54:38 -04:00
Stella Biderman
501ff0c963
Add gpu-calc command and document precision/quantization options
...
New `obliteratus gpu-calc` subcommand estimates minimum GPU count from
model params, dtype, and GPU VRAM. Auto-detects param counts from HF
configs including MoE expert structure.
README now covers --dtype, --quantization flags, the gpu-calc command,
and references both in the "Choosing the right setup" table.
2026-03-17 14:01:18 -04:00
Stella Biderman
79b469d3dc
Add DeepSeek-R1-Distill-Llama-70B pipeline parallel benchmarks
...
Benchmarked 70B dense model (149 GB bf16) on 2/3/4/8 A100-80GB GPUs.
3 GPUs was fastest (536s), confirming minimum-viable-GPU-count guidance.
Combined stage breakdown table for both models.
2026-03-16 15:29:57 -04:00
Stella Biderman
c723da02c8
Document multi-GPU parallelism, benchmarks, and remote SSH execution
...
Add a comprehensive section covering:
- How model sharding (pipeline parallelism) works and its limitations
- GPU selection via --gpus flag
- Pipeline parallel benchmarks on GPT-OSS-120B across 3-8 A100-80GB GPUs
- Stage-by-stage timing breakdown
- When data parallelism helps (and when it doesn't)
- Remote SSH execution with CLI and YAML examples
- Decision table for choosing the right setup
2026-03-16 14:39:22 -04:00
Stella Biderman
a2bb748f1b
Revert "Add data parallel support for PROBE stage"
...
This reverts commit 1a6e2577bb
2026-03-13 16:54:31 -04:00
Stella Biderman
1a6e2577bb
Add data parallel support for PROBE stage
...
When --data-parallel is passed and the model fits on a single GPU,
wraps it with nn.DataParallel to split prompt batches across all
available GPUs during activation collection. Batch size scales by
GPU count. Hooks already move activations to CPU so they work
correctly across replicas.
2026-03-13 01:24:31 -04:00
pliny
984ce14059
Add files via upload
2026-03-05 10:03:46 -08:00
pliny
66ea4a6f86
Add files via upload
2026-03-05 00:50:44 -08:00
pliny
f67f13ca57
Update README.md
2026-03-04 13:35:24 -08:00
pliny
904092fcdb
Update README.md
2026-03-04 12:44:17 -08:00
pliny
0f6114fe87
Add files via upload
2026-03-04 12:38:18 -08:00
Joseph Magly