updated with 2023 material

05_aiTestbed/README.md

@@ -1,56 +1,3 @@
-# BERT (language model) is selected on hands-on section
-
-Bidirectional Encoder Representations from Transformers (BERT) is a transformer-based machine learning technique for natural language processing (NLP) pre-training developed by Google.
-
-
-# SambaNova
-
-1. Login to SN:  
-```
-ssh [email protected] 
-ssh sm-01 (or sm-02)
-```
-
-2. SDK setup:  
-```
-source /software/sambanova/envs/sn_env.sh
-```
-
-3. Copy scripts:  
-```
-cp /var/tmp/Additional/slurm/Models/ANL_Acceptance_RC1_11_5/bert_train-inf.sh ~/
-```
-
-4. Run scripts:  
-```
-cd ~
-./bert_train-inf.sh
-```
-
-
-# Cerebras
-
-1. Login to CS-2:  
-```
-ssh [email protected] 
-ssh cs2-01-med1
-```
-
-2. Copy scripts:  
-```
-cp -r /software/cerebras/model_zoo ~/  
-cd model_zoo/modelzoo/transformers/tf/bert  
-```
-Ignore any permissions errors during the copy of the subdirectory `modelzoo-R1.3.0_2/`.
-
-
-Next, modify `data_dir` to `'/software/cerebras/dataset/bert_large/msl128/'` in `configs/params_bert_large_msl128.yaml`, **in two places**.
-
-4. Run scripts:  
-```
-MODELDIR=model_dir_bert_large_msl128_$(hostname)  
-rm -r $MODELDIR  
-time -p csrun_cpu python run.py --mode=train --compile_only --params configs/params_bert_large_msl128.yaml --model_dir $MODELDIR --cs_ip $CS_IP  
-time -p csrun_wse python run.py --mode=train --params configs/params_bert_large_msl128.yaml --model_dir $MODELDIR --cs_ip $CS_IP
-```
+# Introduction to AI Testbeds at ALCF and hands
 
+Please refer to [Slides here](./AI%20Testbeds%20Hands-on%20ATPESC2023.pdf) and [online documentation](https://docs.alcf.anl.gov/ai-testbed/getting-started/)