scdl README fixes

Author: polinabinder1

sub-packages/bionemo-scdl/README.md

@@ -167,11 +167,13 @@ The runtime is examined on the Tahoe 100M dataset, which containes over 100 mill
 
 ![Throughput](https://raw.githubusercontent.com/NVIDIA/bionemo-framework/pbinder/scdl_add_to_edawson/sub-packages/bionemo-scdl/assets/tahoe_throughput.png)
 
-### Using Neighbor Information in Single Cell Datasets
+To replicate this on your machine, see: [Tahoe 100M Profiling](https://github.com/NVIDIA/bionemo-framework/blob/main/sub-packages/bionemo-scdl/simple-benchmark/README.md#tahoe-100m) section.
+
+## Using Neighbor Information in Single Cell Datasets
 
 SCDL now supports loading and utilizing neighbor information from AnnData objects. This is particularly useful for tasks that require knowledge of cell neighborhoods, trajectory analysis, or spatial relationships.
 
-#### Neighbor Data Structure in AnnData
+### Neighbor Data Structure in AnnData
 
 The neighbor functionality reads neighbor information from the **`.obsp` (observations pairwise) attribute** of the AnnData object and **converts it from sparse matrix format into SCDL's memory-mapped format** for efficient access:
 

sub-packages/bionemo-scdl/simple-benchmark/README.md

@@ -246,6 +246,12 @@ on your system. The following command will run the speedtest on the first plate,
 python scdl_speedtest.py --generate-baseline -i tahoe-100m/h5ad/plate1_filt_Vevo_Tahoe100M_WServicesFrom_ParseGigalab.h5ad
 ```
 
+Alternatively, on the fully converted data:
+
+```bash
+python -m bionemo.scdl.simple_benchmark.scdl_speedtest.py --generate-baseline -i <path to Tahoe 100M in h5ad format> --scdl-path <path to Tahoe 100M in SCDL format> --warmup-time 30 --max-time 120
+```
+
 ## Support
 
 For support, please [file an issue in the BioNeMo Framework GitHub repository](https://github.com/NVIDIA/bionemo-framework/issues).