Upload dataset

README.md

@@ -1,51 +1,39 @@
 ---
 dataset_info:
   features:
-  - name: run_name
-    dtype: string
-  - name: params
+  - name: depth
     dtype: int64
   - name: width
     dtype: int64
-  - name: depth
-    dtype: int64
   - name: tokens
     dtype: int64
-  - name: step
-    dtype: int64
-  - name: num_nodes
-    dtype: int64
-  - name: seconds_per_step
+  - name: FLOPs_per_token
     dtype: float64
-  - name: final_loss
-    dtype: float64
-  - name: legend_name
-    dtype: string
-  - name: flops_per_tok_col
-    dtype: int64
-  - name: flops_per_tok_me
+  - name: FLOPs
     dtype: float64
-  - name: flops_ratio
+  - name: params
     dtype: float64
-  - name: flops_per_token
+  - name: params_with_embeds
     dtype: float64
-  - name: FLOPs
+  - name: FLOPs_6N
     dtype: float64
-  - name: GPU_hours
+  - name: params_pred_loss
     dtype: float64
-  - name: aspect_ratio
+  - name: wd_ratio
     dtype: float64
-  - name: tokens_per_param
+  - name: wd_pred_loss
     dtype: float64
+  - name: bucket
+    dtype: string
   splits:
   - name: train
-    num_bytes: 3491
-    num_examples: 21
-  download_size: 9552
-  dataset_size: 3491
+    num_bytes: 1772
+    num_examples: 13
+  download_size: 6824
+  dataset_size: 1772
 configs:
 - config_name: default
   data_files:
   - split: train
-    path: data_cache/figure_2_flops_mins_hot/train-*
+    path: data_cache/mins_hot_100b+_width_depth_params_relaxed_params/train-*
 ---

data_cache/mins_hot_100b+_width_depth_params_relaxed_params/train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38e4abba049bd2d3f5e2d8fcf02265a9d624b586f75783bae87bca17326b2de4
+size 6824