Issues
- 0
- 0
Handle CUDA Device side assert
#36 opened by JadenFiotto-Kaufman - 0
Documentation + README
#26 opened by JadenFiotto-Kaufman - 0
Limit NDIF usage
#34 opened by JadenFiotto-Kaufman - 0
- 0
Use /stats endpoint to check which models are running in the "request" service
#32 opened by JadenFiotto-Kaufman - 0
Make API key functionality optional
#30 opened by JadenFiotto-Kaufman - 0
Allow specification of nnsight object init kwargs via "model" service environment variable
#29 opened by JadenFiotto-Kaufman - 0
Allow specification of nnsight object type via "model" service environment variable
#28 opened by JadenFiotto-Kaufman - 0
Log API key along with job ID
#27 opened by JadenFiotto-Kaufman - 0
Ensure un-pickled tensors go to cpu.
#24 opened by JadenFiotto-Kaufman - 0
Check what happens when you set the value of early layer with late layer.
#20 opened by JadenFiotto-Kaufman - 0
- 0
- 0
Add int arg to invoker.next()
#21 opened by JadenFiotto-Kaufman - 0
Use Pydantic Validators
#18 opened by JadenFiotto-Kaufman - 0
- 0
Ad hoc module handling
#19 opened by JadenFiotto-Kaufman - 0
- 0
We have tree.model in Intervention now, use it to set device in prepare inputs. Also make IntTree.model = None in init
#10 opened by JadenFiotto-Kaufman - 0
Add logging for creation of proxy
#9 opened by JadenFiotto-Kaufman - 0
- 0
Add InterventionTree Class
#7 opened by JadenFiotto-Kaufman - 0
Merge in torchfx
#8 opened by JadenFiotto-Kaufman