{
  "schema_version": "1.0.0",
  "source_workbook": "Modelling.xlsx",
  "source_sheet": "DefaultTrainingParameters",
  "generated_at": "2026-03-25",
  "records": [
    {
      "basemodel": "prokbert-mini",
      "learning_rate": 0.0004,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 256,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 256,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini"
    },
    {
      "basemodel": "prokbert-mini",
      "learning_rate": 0.0004,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 512,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 128,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini"
    },
    {
      "basemodel": "prokbert-mini",
      "learning_rate": 0.0004,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 1024,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 80,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini"
    },
    {
      "basemodel": "prokbert-mini",
      "learning_rate": 0.0002,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 2048,
      "max_token_length": 2048,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 32,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini"
    },
    {
      "basemodel": "prokbert-mini-long",
      "learning_rate": 0.0001,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 128,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 768,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini-long"
    },
    {
      "basemodel": "prokbert-mini-long",
      "learning_rate": 0.0001,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 256,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 384,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini-long"
    },
    {
      "basemodel": "prokbert-mini-long",
      "learning_rate": 0.0001,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 512,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 128,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini-long"
    },
    {
      "basemodel": "prokbert-mini-long",
      "learning_rate": 5e-05,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 1536,
      "max_token_length": 786,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 128,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini-long"
    },
    {
      "basemodel": "prokbert-mini-long",
      "learning_rate": 5e-05,
      "learning_type": "linear",
      "seq_length_min": 1537,
      "seq_length_max": 2048,
      "max_token_length": 1024,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 64,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini-long"
    },
    {
      "basemodel": "prokbert-mini-long",
      "learning_rate": 0.0001,
      "learning_type": "linear",
      "seq_length_min": 2049,
      "seq_length_max": 4096,
      "max_token_length": 2048,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 32,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini-long"
    },
    {
      "basemodel": "prokbert-mini-c",
      "learning_rate": 0.0004,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 256,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 512,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini-c"
    },
    {
      "basemodel": "prokbert-mini-c",
      "learning_rate": 0.0004,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 512,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 256,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini-c"
    },
    {
      "basemodel": "prokbert-mini-c",
      "learning_rate": 0.0004,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 1024,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 128,
      "task_type": "sequence_classification",
      "model_id": "prokbert-mini-c"
    },
    {
      "basemodel": "nucleotide-transformer-v2-50m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 50,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 640,
      "task_type": "sequence_classification",
      "model_id": "nt50"
    },
    {
      "basemodel": "nucleotide-transformer-v2-50m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 100,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 320,
      "task_type": "sequence_classification",
      "model_id": "nt50"
    },
    {
      "basemodel": "nucleotide-transformer-v2-50m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 200,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 32,
      "task_type": "sequence_classification",
      "model_id": "nt50"
    },
    {
      "basemodel": "nucleotide-transformer-v2-50m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 1536,
      "max_token_length": 300,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 16,
      "task_type": "sequence_classification",
      "model_id": "nt50"
    },
    {
      "basemodel": "nucleotide-transformer-v2-50m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 1537,
      "seq_length_max": 2048,
      "max_token_length": 400,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 8,
      "task_type": "sequence_classification",
      "model_id": "nt50"
    },
    {
      "basemodel": "nucleotide-transformer-v2-50m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 2049,
      "seq_length_max": 4096,
      "max_token_length": 500,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 4,
      "task_type": "sequence_classification",
      "model_id": "nt50"
    },
    {
      "basemodel": "nucleotide-transformer-v2-100m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 50,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 1024,
      "task_type": "sequence_classification",
      "model_id": "nt100"
    },
    {
      "basemodel": "nucleotide-transformer-v2-100m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 100,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 512,
      "task_type": "sequence_classification",
      "model_id": "nt100"
    },
    {
      "basemodel": "nucleotide-transformer-v2-100m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 200,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 384,
      "task_type": "sequence_classification",
      "model_id": "nt100"
    },
    {
      "basemodel": "nucleotide-transformer-v2-100m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 1536,
      "max_token_length": 300,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 256,
      "task_type": "sequence_classification",
      "model_id": "nt100"
    },
    {
      "basemodel": "nucleotide-transformer-v2-100m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 1537,
      "seq_length_max": 2048,
      "max_token_length": 400,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 128,
      "task_type": "sequence_classification",
      "model_id": "nt100"
    },
    {
      "basemodel": "nucleotide-transformer-v2-100m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 2049,
      "seq_length_max": 4096,
      "max_token_length": 500,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 128,
      "task_type": "sequence_classification",
      "model_id": "nt100"
    },
    {
      "basemodel": "nucleotide-transformer-v2-250m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 50,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 256,
      "task_type": "sequence_classification",
      "model_id": "nt250"
    },
    {
      "basemodel": "nucleotide-transformer-v2-250m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 100,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 128,
      "task_type": "sequence_classification",
      "model_id": "nt250"
    },
    {
      "basemodel": "nucleotide-transformer-v2-250m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 200,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 2,
      "batch_size": 64,
      "task_type": "sequence_classification",
      "model_id": "nt250"
    },
    {
      "basemodel": "nucleotide-transformer-v2-250m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 1536,
      "max_token_length": 300,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 4,
      "batch_size": 64,
      "task_type": "sequence_classification",
      "model_id": "nt250"
    },
    {
      "basemodel": "nucleotide-transformer-v2-250m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 1537,
      "seq_length_max": 2048,
      "max_token_length": 400,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 4,
      "batch_size": 64,
      "task_type": "sequence_classification",
      "model_id": "nt250"
    },
    {
      "basemodel": "nucleotide-transformer-v2-250m-multi-species",
      "learning_rate": 2e-05,
      "learning_type": "linear",
      "seq_length_min": 2049,
      "seq_length_max": 4096,
      "max_token_length": 500,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 4,
      "batch_size": 8,
      "task_type": "sequence_classification",
      "model_id": "nt250"
    },
    {
      "basemodel": "nucleotide-transformer-v2-500m-multi-species",
      "learning_rate": 5e-06,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 50,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 32,
      "task_type": "sequence_classification",
      "model_id": "nt500"
    },
    {
      "basemodel": "nucleotide-transformer-v2-500m-multi-species",
      "learning_rate": 5e-06,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 100,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 16,
      "task_type": "sequence_classification",
      "model_id": "nt500"
    },
    {
      "basemodel": "nucleotide-transformer-v2-500m-multi-species",
      "learning_rate": 5e-06,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 200,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 8,
      "task_type": "sequence_classification",
      "model_id": "nt500"
    },
    {
      "basemodel": "nucleotide-transformer-v2-500m-multi-species",
      "learning_rate": 5e-06,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 1536,
      "max_token_length": 300,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 4,
      "task_type": "sequence_classification",
      "model_id": "nt500"
    },
    {
      "basemodel": "nucleotide-transformer-v2-500m-multi-species",
      "learning_rate": 5e-06,
      "learning_type": "linear",
      "seq_length_min": 1537,
      "seq_length_max": 2048,
      "max_token_length": 400,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 2,
      "task_type": "sequence_classification",
      "model_id": "nt500"
    },
    {
      "basemodel": "nucleotide-transformer-v2-500m-multi-species",
      "learning_rate": 5e-06,
      "learning_type": "linear",
      "seq_length_min": 2049,
      "seq_length_max": 4096,
      "max_token_length": 500,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 4,
      "task_type": "sequence_classification",
      "model_id": "nt500"
    },
    {
      "basemodel": "nucleotide-transformer-2.5b-multi-species",
      "learning_rate": 1e-05,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 50,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 64,
      "task_type": "sequence_classification",
      "model_id": "nt2.5b"
    },
    {
      "basemodel": "nucleotide-transformer-2.5b-multi-species",
      "learning_rate": 1e-05,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 100,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 16,
      "task_type": "sequence_classification",
      "model_id": "nt2.5b"
    },
    {
      "basemodel": "nucleotide-transformer-2.5b-multi-species",
      "learning_rate": 1e-05,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 200,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 8,
      "task_type": "sequence_classification",
      "model_id": "nt2.5b"
    },
    {
      "basemodel": "nucleotide-transformer-2.5b-multi-species",
      "learning_rate": 1e-05,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 1536,
      "max_token_length": 300,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 4,
      "task_type": "sequence_classification",
      "model_id": "nt2.5b"
    },
    {
      "basemodel": "nucleotide-transformer-2.5b-multi-species",
      "learning_rate": 1e-05,
      "learning_type": "linear",
      "seq_length_min": 1537,
      "seq_length_max": 2048,
      "max_token_length": 400,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 2,
      "task_type": "sequence_classification",
      "model_id": "nt2.5b"
    },
    {
      "basemodel": "nucleotide-transformer-2.5b-multi-species",
      "learning_rate": 1e-05,
      "learning_type": "linear",
      "seq_length_min": 2049,
      "seq_length_max": 4096,
      "max_token_length": 500,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 1,
      "task_type": "sequence_classification",
      "model_id": "nt2.5b"
    },
    {
      "basemodel": "DNABERT-2-117M",
      "learning_rate": 0.0005,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 70,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 74,
      "task_type": "sequence_classification",
      "model_id": "dnabert2"
    },
    {
      "basemodel": "DNABERT-2-117M",
      "learning_rate": 0.0005,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 150,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 75,
      "task_type": "sequence_classification",
      "model_id": "dnabert2"
    },
    {
      "basemodel": "DNABERT-2-117M",
      "learning_rate": 0.0005,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 256,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 76,
      "task_type": "sequence_classification",
      "model_id": "dnabert2"
    },
    {
      "basemodel": "DNABERT-2-117M",
      "learning_rate": 0.0005,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 1536,
      "max_token_length": 384,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 77,
      "task_type": "sequence_classification",
      "model_id": "dnabert2"
    },
    {
      "basemodel": "DNABERT-2-117M",
      "learning_rate": 0.0005,
      "learning_type": "linear",
      "seq_length_min": 1537,
      "seq_length_max": 2048,
      "max_token_length": 512,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 2,
      "batch_size": 78,
      "task_type": "sequence_classification",
      "model_id": "dnabert2"
    },
    {
      "basemodel": "DNABERT-2-117M",
      "learning_rate": 0.0005,
      "learning_type": "linear",
      "seq_length_min": 2049,
      "seq_length_max": 4096,
      "max_token_length": 1024,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 2,
      "batch_size": 79,
      "task_type": "sequence_classification",
      "model_id": "dnabert2"
    },
    {
      "basemodel": "METAGENE-1",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 85,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 1,
      "batch_size": 80,
      "task_type": "sequence_classification",
      "model_id": "metagene1"
    },
    {
      "basemodel": "METAGENE-1",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 170,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 2,
      "batch_size": 81,
      "task_type": "sequence_classification",
      "model_id": "metagene1"
    },
    {
      "basemodel": "METAGENE-1",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 340,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 4,
      "batch_size": 82,
      "task_type": "sequence_classification",
      "model_id": "metagene1"
    },
    {
      "basemodel": "METAGENE-1",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 1536,
      "max_token_length": 510,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 4,
      "batch_size": 83,
      "task_type": "sequence_classification",
      "model_id": "metagene1"
    },
    {
      "basemodel": "METAGENE-1",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 1537,
      "seq_length_max": 2048,
      "max_token_length": 640,
      "dtype": "bf16",
      "gpu_memory": 40,
      "system": "komondor",
      "gradient_accumulation_steps": 4,
      "batch_size": 84,
      "task_type": "sequence_classification",
      "model_id": "metagene1"
    },
    {
      "basemodel": "evo-1-8k-base",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 0,
      "seq_length_max": 256,
      "max_token_length": 258,
      "dtype": "bf16",
      "gpu_memory": 64,
      "system": "Leonardo",
      "gradient_accumulation_steps": 1,
      "batch_size": 85,
      "task_type": "sequence_classification",
      "model_id": "evo1-8k"
    },
    {
      "basemodel": "evo-1-8k-base",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 257,
      "seq_length_max": 512,
      "max_token_length": 514,
      "dtype": "bf16",
      "gpu_memory": 32,
      "system": "Leonardo",
      "gradient_accumulation_steps": 2,
      "batch_size": 86,
      "task_type": "sequence_classification",
      "model_id": "evo1-8k"
    },
    {
      "basemodel": "evo-1-8k-base",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 513,
      "seq_length_max": 1024,
      "max_token_length": 1026,
      "dtype": "bf16",
      "gpu_memory": 16,
      "system": "Leonardo",
      "gradient_accumulation_steps": 4,
      "batch_size": 87,
      "task_type": "sequence_classification",
      "model_id": "evo1-8k"
    },
    {
      "basemodel": "evo-1-8k-base",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 1025,
      "seq_length_max": 1536,
      "max_token_length": 1538,
      "dtype": "bf16",
      "gpu_memory": 8,
      "system": "Leonardo",
      "gradient_accumulation_steps": 8,
      "batch_size": 88,
      "task_type": "sequence_classification",
      "model_id": "evo1-8k"
    },
    {
      "basemodel": "evo-1-8k-base",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 1537,
      "seq_length_max": 2048,
      "max_token_length": 2050,
      "dtype": "bf16",
      "gpu_memory": 8,
      "system": "Leonardo",
      "gradient_accumulation_steps": 16,
      "batch_size": 89,
      "task_type": "sequence_classification",
      "model_id": "evo1-8k"
    },
    {
      "basemodel": "evo-1-8k-base",
      "learning_rate": 0.0006,
      "learning_type": "linear",
      "seq_length_min": 2049,
      "seq_length_max": 4096,
      "max_token_length": 4100,
      "dtype": "bf16",
      "gpu_memory": 4,
      "system": "Leonardo",
      "gradient_accumulation_steps": 16,
      "batch_size": 90,
      "task_type": "sequence_classification",
      "model_id": "evo1-8k"
    }
  ]
}