Merge pull request #138 from unslothai/fix/reset-max-steps-epoch-defaults

Override Model Defaults for num_epochs and max_steps
2026-02-17 22:51:40 +04:00 · 2026-02-17 22:51:40 +04:00 · c13e1594de
commit c13e1594de
parent 7874fa9066 299bc65e36
63 changed files with 252 additions and 189 deletions
--- a/studio/backend/assets/configs/model_defaults/default.yaml
+++ b/studio/backend/assets/configs/model_defaults/default.yaml
@ -3,13 +3,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 5e-5
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_ratio: 0.1
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.01
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/ernie/unsloth_ERNIE-4.5-21B-A3B-PT.yaml
+++ b/studio/backend/assets/configs/model_defaults/ernie/unsloth_ERNIE-4.5-21B-A3B-PT.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 4
  gradient_accumulation_steps: 2
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/ernie/unsloth_ERNIE-4.5-VL-28B-A3B-PT.yaml
+++ b/studio/backend/assets/configs/model_defaults/ernie/unsloth_ERNIE-4.5-VL-28B-A3B-PT.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 2
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/falcon/tiiuae_Falcon-H1-0.5B-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/falcon/tiiuae_Falcon-H1-0.5B-Instruct.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 8
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.01
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_codegemma-7b-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_codegemma-7b-bnb-4bit.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 4096
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_functiongemma-270m-it.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_functiongemma-270m-it.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 4096
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 4
  gradient_accumulation_steps: 2
  warmup_steps: 10
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-2-27b-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-2-27b-bnb-4bit.yaml
@ -3,13 +3,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-2-2b.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-2-2b.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.01
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-270m-it.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-270m-it.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 5e-5
  batch_size: 4
  gradient_accumulation_steps: 1
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-27b-it.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-27b-it.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-4b-it.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-4b-it.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-4b-pt.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-4b-pt.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 2
+  # num_epochs: 2
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_ratio: 0.03
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3n-E4B-it.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3n-E4B-it.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 1024
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3n-E4B.yaml
+++ b/studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3n-E4B.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 2
+  # num_epochs: 2
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_ratio: 0.03
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gpt-oss/unsloth_gpt-oss-120b.yaml
+++ b/studio/backend/assets/configs/model_defaults/gpt-oss/unsloth_gpt-oss-120b.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 4096
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 4
  gradient_accumulation_steps: 1
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/gpt-oss/unsloth_gpt-oss-20b.yaml
+++ b/studio/backend/assets/configs/model_defaults/gpt-oss/unsloth_gpt-oss-20b.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 1024
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/granite/unsloth_granite-4.0-350m-unsloth-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/granite/unsloth_granite-4.0-350m-unsloth-bnb-4bit.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/granite/unsloth_granite-4.0-h-micro.yaml
+++ b/studio/backend/assets/configs/model_defaults/granite/unsloth_granite-4.0-h-micro.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/llama/unsloth_Llama-3.2-11B-Vision-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/llama/unsloth_Llama-3.2-11B-Vision-Instruct.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/llama/unsloth_Llama-3.2-1B-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/llama/unsloth_Llama-3.2-1B-Instruct.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 5
+  # num_epochs: 5
+  num_epochs: 0
  learning_rate: 2e-5
  batch_size: 1
  gradient_accumulation_steps: 8
  warmup_steps: 0
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.01
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/llama/unsloth_Llama-3.2-3B-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/llama/unsloth_Llama-3.2-3B-Instruct.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/llama/unsloth_Llama-3.3-70B-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/llama/unsloth_Llama-3.3-70B-Instruct.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/llama/unsloth_Meta-Llama-3.1-70B-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/llama/unsloth_Meta-Llama-3.1-70B-bnb-4bit.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/llama/unsloth_Meta-Llama-3.1-8B-Instruct-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/llama/unsloth_Meta-Llama-3.1-8B-Instruct-bnb-4bit.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 8192
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/llama/unsloth_llama-3-8b-Instruct-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/llama/unsloth_llama-3-8b-Instruct-bnb-4bit.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/llama/unsloth_llama-3-8b-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/llama/unsloth_llama-3-8b-bnb-4bit.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/llasa/unsloth_Llasa-3B.yaml
+++ b/studio/backend/assets/configs/model_defaults/llasa/unsloth_Llasa-3B.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 5e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/mistral/unsloth_Magistral-Small-2509-unsloth-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/mistral/unsloth_Magistral-Small-2509-unsloth-bnb-4bit.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 2
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/mistral/unsloth_Ministral-3-3B-Instruct-2512.yaml
+++ b/studio/backend/assets/configs/model_defaults/mistral/unsloth_Ministral-3-3B-Instruct-2512.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 4
  gradient_accumulation_steps: 2
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/mistral/unsloth_Mistral-Nemo-Base-2407-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/mistral/unsloth_Mistral-Nemo-Base-2407-bnb-4bit.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/mistral/unsloth_Mistral-Small-Instruct-2409.yaml
+++ b/studio/backend/assets/configs/model_defaults/mistral/unsloth_Mistral-Small-Instruct-2409.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/mistral/unsloth_Pixtral-12B-2409.yaml
+++ b/studio/backend/assets/configs/model_defaults/mistral/unsloth_Pixtral-12B-2409.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/mistral/unsloth_mistral-7b-instruct-v0.3-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/mistral/unsloth_mistral-7b-instruct-v0.3-bnb-4bit.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/mistral/unsloth_mistral-7b-v0.3-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/mistral/unsloth_mistral-7b-v0.3-bnb-4bit.yaml
@ -3,13 +3,14 @@
 # Also applies to: "unsloth/mistral-7b-v0.3", "mistralai/Mistral-7B-v0.3",
 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/other/OuteAI_Llama-OuteTTS-1.0-1B.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/OuteAI_Llama-OuteTTS-1.0-1B.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/other/Spark-TTS-0.5B_LLM.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/Spark-TTS-0.5B_LLM.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/other/sesame_csm-1b.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/sesame_csm-1b.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/other/unsloth_LFM2-1.2B.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/unsloth_LFM2-1.2B.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/other/unsloth_Nemotron-3-Nano-30B-A3B.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/unsloth_Nemotron-3-Nano-30B-A3B.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 4
  gradient_accumulation_steps: 2
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/other/unsloth_PaddleOCR-VL.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/unsloth_PaddleOCR-VL.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 5e-5
  batch_size: 4
  gradient_accumulation_steps: 2
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/other/unsloth_answerdotai_ModernBERT-large.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/unsloth_answerdotai_ModernBERT-large.yaml
@ -3,13 +3,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 1
+  # num_epochs: 1
+  num_epochs: 0
  learning_rate: 5e-5
  batch_size: 32
  gradient_accumulation_steps: 1
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/other/unsloth_orpheus-3b-0.1-ft.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/unsloth_orpheus-3b-0.1-ft.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/other/unsloth_tinyllama-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/unsloth_tinyllama-bnb-4bit.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 4096
-  num_epochs: 1
+  # num_epochs: 1
+  num_epochs: 0
  learning_rate: 2e-5
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_ratio: 0.1
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.1
  random_seed: 3407
  packing: true
--- a/studio/backend/assets/configs/model_defaults/other/unsloth_whisper-large-v3.yaml
+++ b/studio/backend/assets/configs/model_defaults/other/unsloth_whisper-large-v3.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 448
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 1e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/phi/unsloth_Phi-3-medium-4k-instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/phi/unsloth_Phi-3-medium-4k-instruct.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/phi/unsloth_Phi-3.5-mini-instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/phi/unsloth_Phi-3.5-mini-instruct.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/phi/unsloth_Phi-4-14B-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/phi/unsloth_Phi-4-14B-Instruct.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/phi/unsloth_Phi-4.yaml
+++ b/studio/backend/assets/configs/model_defaults/phi/unsloth_Phi-4.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2-7B.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2-7B.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2-VL-7B-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2-VL-7B-Instruct.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-1.5B-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-1.5B-Instruct.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 4096
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 1e-5
  batch_size: 1
  gradient_accumulation_steps: 64
  warmup_ratio: 0.1
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 42
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-7B.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-7B.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-Coder-1.5B-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-Coder-1.5B-Instruct.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-Coder-14B-Instruct.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-Coder-14B-Instruct.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 1
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-Coder-7B-Instruct-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-Coder-7B-Instruct-bnb-4bit.yaml
@ -4,13 +4,14 @@

 training:
  max_seq_length: 32768
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-VL-7B-Instruct-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen2.5-VL-7B-Instruct-bnb-4bit.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-0.6B.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-0.6B.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 1024
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 5e-5
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-14B-Base-unsloth-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-14B-Base-unsloth-bnb-4bit.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-14B.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-14B.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-32B.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-32B.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-4B-Instruct-2507.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-4B-Instruct-2507.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-4B-Thinking-2507.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-4B-Thinking-2507.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false
--- a/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-VL-8B-Instruct-unsloth-bnb-4bit.yaml
+++ b/studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-VL-8B-Instruct-unsloth-bnb-4bit.yaml
@ -5,13 +5,14 @@

 training:
  max_seq_length: 2048
-  num_epochs: 4
+  # num_epochs: 4
+  num_epochs: 0
  learning_rate: 2e-4
  batch_size: 2
  gradient_accumulation_steps: 4
  warmup_steps: 5
-  max_steps: 0
-  save_steps: 0
+  max_steps: 30
+  save_steps: 30
  weight_decay: 0.001
  random_seed: 3407
  packing: false