Spaces:

ProCreations
/

repro-nanoquant-efficient-sub-1-bit-quantization-of-large-language-models

Running

repro-nanoquant-efficient-sub-1-bit-quantization-of-large-language-models

476 kB

Ctrl+K

1 contributor

Add the missing Claim 5 page: rerun the COMPAS case study on real ProPublica data; optimal fair thresholds differ by 0.08-0.13 across groups and forcing a single threshold costs 52-341 utility

0690c42 verified 5 days ago

code
Add the missing Claim 5 page: rerun the COMPAS case study on real ProPublica data; optimal fair thresholds differ by 0.08-0.13 across groups and forcing a single threshold costs 52-341 utility 5 days ago
outputs
Map exact current anchored claims to existing independent evidence 8 days ago
pages
Add the missing Claim 5 page: rerun the COMPAS case study on real ProPublica data; optimal fair thresholds differ by 0.08-0.13 across groups and forcing a single threshold costs 52-341 utility 5 days ago
.gitattributes

1.74 kB
Map exact current anchored claims to existing independent evidence 8 days ago
BUNDLE_SHA256SUMS.txt

2.67 kB
Map exact current anchored claims to existing independent evidence 8 days ago
README.md

519 Bytes
Map exact current anchored claims to existing independent evidence 8 days ago
anchored_claim_map.json

3.49 kB
Map exact current anchored claims to existing independent evidence 8 days ago
audit_fair_calibrated.py

13.9 kB
Map exact current anchored claims to existing independent evidence 8 days ago
boundary_trace.py

10.6 kB
Map exact current anchored claims to existing independent evidence 8 days ago
bucket-icon.svg

413 Bytes
Update logbook: Repro - NanoQuant: Efficient Sub-1-bit Quantization of Large Language Models 15 days ago
index.html

1.84 kB
Map exact current anchored claims to existing independent evidence 8 days ago
logbook.css

29.6 kB
Update logbook: Repro - NanoQuant: Efficient Sub-1-bit Quantization of Large Language Models 15 days ago
logbook.js

77.3 kB
Update logbook: Repro - NanoQuant: Efficient Sub-1-bit Quantization of Large Language Models 15 days ago
logbook.json

2.63 kB
Map exact current anchored claims to existing independent evidence 8 days ago
official_claims.json

1.04 kB
Map exact current anchored claims to existing independent evidence 8 days ago
outputs_compas_results.json

3.18 kB
Add the missing Claim 5 page: rerun the COMPAS case study on real ProPublica data; optimal fair thresholds differ by 0.08-0.13 across groups and forcing a single threshold costs 52-341 utility 5 days ago
poster_embed.html

1.71 kB
Map exact current anchored claims to existing independent evidence 8 days ago
requirements.txt

46 Bytes
Map exact current anchored claims to existing independent evidence 8 days ago
trackio-logo-light.png

30 kB
Update logbook: Repro - NanoQuant: Efficient Sub-1-bit Quantization of Large Language Models 15 days ago
trackio-logo.png

55.6 kB
Update logbook: Repro - NanoQuant: Efficient Sub-1-bit Quantization of Large Language Models 15 days ago
trackio-wordmark-dark.png

89.8 kB
Update logbook: Repro - NanoQuant: Efficient Sub-1-bit Quantization of Large Language Models 15 days ago