Update accelerate_config_32b.yaml

train_and_evaluate/accelerate_config_32b.yaml

 compute_environment: LOCAL_MACHINE
 debug: false
 deepspeed_config:
-  gradient_accumulation_steps: 32
+  gradient_accumulation_steps: 64
   gradient_clipping: 1.0
-  offload_optimizer_device: none
-  offload_param_device: none
-  zero3_init_flag: true
-  zero3_save_16bit_model: true
+  offload_optimizer_device: 'cpu'
+  offload_param_device: 'cpu'
+  zero3_init_flag: false
   zero_stage: 3
+  zero3_save_16bit_model: true
 distributed_type: DEEPSPEED
-downcast_bf16: 'no'
+downcast_bf16: 'true'
 enable_cpu_affinity: false
 machine_rank: 0
 main_training_function: main