cURL
curl --request POST \ --url https://training-suite.simplismart.ai/api/llm/train/ \ --header 'Authorization: Bearer <token>' \ --header 'Content-Type: multipart/form-data' \ --form 'training_config_json={ "model_type": "vlm", "base_model": "AIDC-AI/Ovis1.6-Llama3.2-3B", "adapter": { "lora_r": 16 }, "quantization": { "bits": 4, "llm_int8_threshold": 6, "llm_int8_has_fp16_weight": false, "bnb_4bit_compute_dtype": "float16", "bnb_4bit_use_double_quant": true, "bnb_4bit_quant_type": "nf4" }, "trainer": { "type": "finetune", "learning_rate": 0.0001, "batch_size": 1, "epochs": 1, "gradient_accumulation_steps": 16, "logging_steps": 1 }, "image_column": "url", "output_column": "caption", "prompt": { "template": "Could you please interpret the image and write a detailed caption?" } } ' \ --form experiment_name=my_first_vlm_job \ --form org=0bf00b43-430a-4ca3-a8b3-b13cc8dc6d4d
{ "request_id": "<string>", "status": "<string>", "message": "<string>" }
Submit a new training job with the specified configuration, training data, and metadata.
JWT token for authentication
Bearer token for authentication and authorization.
Training job submitted successfully.
The response is of type object.
object