Spaces:
Running on CPU Upgrade
Running on CPU Upgrade
Commit History
26 days ago
-
fix TReB QA mislabel and PIQA category; add model-size slide and Q&A notes to COLM spotlight
The treb_qa benchmark (lighteval/treb_table_retrieval) was labeled TriviaQA in the blog and charts; relabel it and cite TReB. The macro score counts PIQA under Reasoning, so the setup list and correlation chart now group it there. Fix the reversed color legend in the score-correlation caption. The COLM deck gets the trained-model-size slide, a training-compute estimate and benchmark cheat sheet in the notes, the 4:30 poster time, and DROP instead of TriviaQA on the trade-off slide. Co-authored-by: Cursor <cursoragent@cursor.com>c24632b
-
add synthetic data proportion experiment
d22ec15
-
added analysis with differently sized student models
9894e4e
-
fix cut off y axis labels
58221a2 -
added speedup annotation for main plot
fe33625
-
improved plots for mobile
7db0003
-
updated synthetic only results
e575fa7
-
improved axis formatting
4affc7e -
fixed issue with average setups producing invalid results
e569197
-
show the average setup by default
4fad4ec
-
prettified experiment plots
fdfc515
-
fix main chart dclm darkmode issue
b23f8a3
-
make plots look nicer
387754d
-
moved the plot more to the right as to not cut off the display names
0d3f430 -
support different setups so that we can reduce the number of charts a bit
c94a38b -
make baselines striped in bar chart and dashed in line chart
8cdc217 -
removed redundant defaultView and added dclm and fw-edu-hq baselines everywhere with stable colors for less mental load
1834e40 -
switched to just one csv file with the complete results
3b14ed8 -
make view switch simpler
ee8663f
-
made plot d3 component more general and added finephrase vs baseline comparison
a6df89a