Spaces:

visualisable-ai
/

api

Paused

App Files Files Community

614 kB

Ctrl+K

Ctrl+K

4 contributors

History: 98 commits

gary-boon

Claude Opus 4.5

Add per-step memory cleanup for large model support

a94eb19 7 months ago

.claude
Add Claude Code configuration 9 months ago
.github
fix: Skip heavy ML deps in CI security checks 7 months ago
backend
Add per-step memory cleanup for large model support 7 months ago
components
Add layer_stride parameter for PromptDiff optimization 10 months ago
docker
Use NGC PyTorch 24.08 for Python 3.10 compatibility 7 months ago
docs
Add Phase 5: Performance optimizations to phased plan 7 months ago
runs
Phase 1: DGX Spark infrastructure 7 months ago
.env.spark.example

772 Bytes
Update .env.spark.example: TORCH_DTYPE now auto-detected 7 months ago
.gitattributes

1.52 kB
initial commit 11 months ago
.gitignore

499 Bytes
Phase 1: DGX Spark infrastructure 7 months ago
Dockerfile

1.14 kB
Fix Dockerfile: add build-essential for numcodecs compilation 7 months ago
FUTURE_CONSIDERATIONS.md

4.11 kB
Add future considerations doc for response size optimization 7 months ago
README.md

909 Bytes
Update HuggingFace Space description 10 months ago
SECURITY.md

4.45 kB
Add GitHub Actions workflow for security scanning and automated deployment 11 months ago
TESTING.md

4.56 kB
Add Code Llama 7B support with hardware-aware filtering and ICL timeout fixes 9 months ago
TEST_RESULTS.md

6.97 kB
Add Code Llama 7B support with hardware-aware filtering and ICL timeout fixes 9 months ago
app.py

1.69 kB
Fix HuggingFace Spaces cache permission error 11 months ago
explore_vocabulary.py

2.51 kB
Add research attention analysis endpoints with Q/K/V extraction 8 months ago
requirements.txt

605 Bytes
Integrate mistral-common for correct Devstral tokenization 7 months ago
setup-api-key.md

1.32 kB
Add API key authentication 11 months ago
test_instrumentation.py

8.99 kB
Add research attention analysis endpoints with Q/K/V extraction 8 months ago
test_multi_model.py

8.61 kB
Add Code Llama 7B support with hardware-aware filtering and ICL timeout fixes 9 months ago