Get Customization Configs

Get a list of available customization configurations and their details.

You can review the returned customization configs to determine which models are available for fine-tuning. These configs are typically added by your cluster administrator during the initial setup of NeMo Customizer.

Prerequisites

Before you can get a list of customization configurations, make sure that you have:

• Access to the NeMo Customizer service via the API or Studio UI.

• Reviewed the customization config reference article.

---

Get a List of Customization Configurations

1. Perform a GET request to the /v1/customization/configs endpoint.

   BASE_MODEL="meta/llama-3.1-8b-instruct"
   FINETUNING_TYPE="lora"
   TRAINING_TYPE="sft"
   BATCH_SIZE="8"
   EPOCHS="1"
   LOG_EVERY_N_STEPS="10"
   DATASET="string"
   
   curl --get \
     "https://${CUSTOMIZER_HOSTNAME}/v1/customization/configs" \
     --data-urlencode "page=1" \
     --data-urlencode "page_size=10" \
     --data-urlencode "sort=-created_at" \
     --data-urlencode "filter[base_model]=${BASE_MODEL}" \
     --data-urlencode "filter[finetuning_type]=${FINETUNING_TYPE}" \
     --data-urlencode "filter[training_type]=${TRAINING_TYPE}" \
     --data-urlencode "filter[batch_size]=${BATCH_SIZE}" \
     --data-urlencode "filter[epochs]=${EPOCHS}" \
     --data-urlencode "filter[log_every_n_steps]=${LOG_EVERY_N_STEPS}" \
     --data-urlencode "filter[dataset]=${DATASET}" \
     --data-urlencode "filter[status]=created" | jq
   

   You can use the name field of a returned config object to set the config parameter when creating a customization job.

2. Review the returned customization configs.

   Example Response

   {
     "object": "list",
     "data": [
       {
         "created_at": "2024-11-26T02:58:55.339737",
         "updated_at": "2024-11-26T02:58:55.339737",
         "name": "customization_config-MedVscVbr4pgLhLgKTLbv9",
         "base_model": "meta/llama-3.1-8b-instruct",
         "training_types": [
           "sft"
         ],
         "finetuning_types": [
           "lora"
         ],
         "precision": "bf16",
         "num_gpus": 2,
         "micro_batch_size": 1,
         "tensor_parallel_size": 1,
         "max_seq_length": 4096,
         "custom_fields": {}
       },
     ],
     "pagination": {
       "page": 1,
       "page_size": 10,
       "current_page_size": 2,
       "total_pages": 1,
       "total_results": 2
     },
     "sort": "created_at"
   }
   

   Tip

   The num_gpus multiplied by the num_nodes is the total number of GPUs required to run a fine tuning.