Tabular Classification
Transformers
Safetensors
felatab
feature-extraction
fela
tabular
in-context-learning
prior-fitted-network
foundation-model
delta-rule
cpu
on-device
custom_code
Eval Results (legacy)
Instructions to use lowdown-labs/fela-tab with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use lowdown-labs/fela-tab with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("lowdown-labs/fela-tab", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
Upload benchmark/tabarena/ec2_run.sh with huggingface_hub
Browse files
benchmark/tabarena/ec2_run.sh
CHANGED
|
@@ -37,6 +37,13 @@ mkdir -p "$WORK/tmp" "$WORK/logs" && cd "$WORK"
|
|
| 37 |
exec 9> "$WORK/.felatab.lock"
|
| 38 |
flock -n 9 || { echo "another felatab run holds the lock; exiting"; exit 0; }
|
| 39 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
upload() { # upload <phase-label>
|
| 41 |
"$WORK/bootvenv/bin/python" - <<EOF
|
| 42 |
from huggingface_hub import HfApi
|
|
|
|
| 37 |
exec 9> "$WORK/.felatab.lock"
|
| 38 |
flock -n 9 || { echo "another felatab run holds the lock; exiting"; exit 0; }
|
| 39 |
|
| 40 |
+
# AutoGluon never cleans its per-task ag-* predictor dirs in TMPDIR (~13G each on
|
| 41 |
+
# whale datasets -> root disk fills and the run crashes with ENOSPC). Prune dirs
|
| 42 |
+
# older than 60 min every 15 min for the life of this box.
|
| 43 |
+
(
|
| 44 |
+
while :; do find "$WORK/tmp" -maxdepth 1 -name 'ag-*' -mmin +60 -exec rm -rf {} + 2>/dev/null; sleep 900; done
|
| 45 |
+
) </dev/null >/dev/null 2>&1 &
|
| 46 |
+
|
| 47 |
upload() { # upload <phase-label>
|
| 48 |
"$WORK/bootvenv/bin/python" - <<EOF
|
| 49 |
from huggingface_hub import HfApi
|