[tiny] Remove deprecated use_async_dataset from configs (#1701)

wizeng23 · web-flow · commit edc471f069bd · 2025-05-28T17:59:21.000-07:00
diff --git a/configs/examples/fineweb_ablation_pretraining/ddp/train.yaml b/configs/examples/fineweb_ablation_pretraining/ddp/train.yaml
@@ -36,7 +36,6 @@ data:
     stream: True
     pack: True
     target_col: "text"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT" # or OUMI
diff --git a/configs/examples/fineweb_ablation_pretraining/fsdp/train.yaml b/configs/examples/fineweb_ablation_pretraining/fsdp/train.yaml
@@ -36,7 +36,6 @@ data:
     stream: True
     pack: True
     target_col: "text"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/projects/coalm/405b_train.yaml b/configs/projects/coalm/405b_train.yaml
@@ -19,15 +19,13 @@ data:
         seed: 42
     collator_name: "text_completions_only_with_padding"
     target_col: "messages"
-    use_async_dataset: True
     seed: 42
   validation:
     datasets:
       - dataset_name: "text_sft_jsonl"
         dataset_path: "/path/to/validation/dataset.jsonl"
     collator_name: "text_completions_only_with_padding"
     target_col: "messages"
-    use_async_dataset: True
     seed: 42
 
 training:
diff --git a/configs/projects/coalm/70b_train.yaml b/configs/projects/coalm/70b_train.yaml
@@ -19,15 +19,13 @@ data:
         seed: 42
     collator_name: "text_completions_only_with_padding"
     target_col: "messages"
-    use_async_dataset: True
     seed: 42
   validation:
     datasets:
       - dataset_name: "text_sft_jsonl"
         dataset_path: "/path/to/validation/dataset.jsonl"
     collator_name: "text_completions_only_with_padding"
     target_col: "messages"
-    use_async_dataset: True
     seed: 42
 
 training:
diff --git a/configs/projects/coalm/8b_train.yaml b/configs/projects/coalm/8b_train.yaml
@@ -19,15 +19,13 @@ data:
         seed: 42
     collator_name: "text_completions_only_with_padding"
     target_col: "messages"
-    use_async_dataset: True
     seed: 42
   validation:
     datasets:
       - dataset_name: "text_sft_jsonl"
         dataset_path: "/path/to/validation/dataset.jsonl"
     collator_name: "text_completions_only_with_padding"
     target_col: "messages"
-    use_async_dataset: True
     seed: 42
 
 training:
diff --git a/configs/recipes/deepseek_r1/sft/distill_llama_70b/full_train.yaml b/configs/recipes/deepseek_r1/sft/distill_llama_70b/full_train.yaml
@@ -26,7 +26,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/deepseek_r1/sft/distill_llama_70b/lora_train.yaml b/configs/recipes/deepseek_r1/sft/distill_llama_70b/lora_train.yaml
@@ -25,7 +25,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/deepseek_r1/sft/distill_llama_70b/qlora_train.yaml b/configs/recipes/deepseek_r1/sft/distill_llama_70b/qlora_train.yaml
@@ -25,7 +25,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/deepseek_r1/sft/distill_llama_8b/full_train.yaml b/configs/recipes/deepseek_r1/sft/distill_llama_8b/full_train.yaml
@@ -27,7 +27,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/deepseek_r1/sft/distill_llama_8b/lora_train.yaml b/configs/recipes/deepseek_r1/sft/distill_llama_8b/lora_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/deepseek_r1/sft/distill_llama_8b/qlora_train.yaml b/configs/recipes/deepseek_r1/sft/distill_llama_8b/qlora_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/deepseek_r1/sft/distill_qwen_1_5b/full_train.yaml b/configs/recipes/deepseek_r1/sft/distill_qwen_1_5b/full_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/deepseek_r1/sft/distill_qwen_1_5b/lora_train.yaml b/configs/recipes/deepseek_r1/sft/distill_qwen_1_5b/lora_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/deepseek_r1/sft/distill_qwen_32b/lora_train.yaml b/configs/recipes/deepseek_r1/sft/distill_qwen_32b/lora_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/pretraining/8b/train.yaml b/configs/recipes/llama3_1/pretraining/8b/train.yaml
@@ -39,7 +39,6 @@ data:
     stream: True
     pack: True
     target_col: "text"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/405b_full/train.yaml b/configs/recipes/llama3_1/sft/405b_full/train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/405b_lora/train.yaml b/configs/recipes/llama3_1/sft/405b_lora/train.yaml
@@ -31,7 +31,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/405b_qlora/train.yaml b/configs/recipes/llama3_1/sft/405b_qlora/train.yaml
@@ -31,7 +31,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/70b_full/train.yaml b/configs/recipes/llama3_1/sft/70b_full/train.yaml
@@ -35,7 +35,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/70b_lora/train.yaml b/configs/recipes/llama3_1/sft/70b_lora/train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/70b_qlora/train.yaml b/configs/recipes/llama3_1/sft/70b_qlora/train.yaml
@@ -30,7 +30,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/8b_full/longctx_train.yaml b/configs/recipes/llama3_1/sft/8b_full/longctx_train.yaml
@@ -50,6 +50,5 @@ data:
         dataset_kwargs:
           seq_length: 32_768
     target_col: "text"
-    use_async_dataset: True
     stream: True
     pack: True
diff --git a/configs/recipes/llama3_1/sft/8b_full/train.yaml b/configs/recipes/llama3_1/sft/8b_full/train.yaml
@@ -31,7 +31,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/8b_lora/fsdp_train.yaml b/configs/recipes/llama3_1/sft/8b_lora/fsdp_train.yaml
@@ -31,7 +31,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/8b_lora/train.yaml b/configs/recipes/llama3_1/sft/8b_lora/train.yaml
@@ -31,7 +31,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_1/sft/8b_qlora/train.yaml b/configs/recipes/llama3_1/sft/8b_qlora/train.yaml
@@ -31,7 +31,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_2/sft/1b_full/train.yaml b/configs/recipes/llama3_2/sft/1b_full/train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_2/sft/3b_full/fsdp_train.yaml b/configs/recipes/llama3_2/sft/3b_full/fsdp_train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_2/sft/3b_full/train.yaml b/configs/recipes/llama3_2/sft/3b_full/train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_2/sft/3b_lora/fsdp_train.yaml b/configs/recipes/llama3_2/sft/3b_lora/fsdp_train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_2/sft/3b_lora/train.yaml b/configs/recipes/llama3_2/sft/3b_lora/train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_2/sft/3b_qlora/fsdp_train.yaml b/configs/recipes/llama3_2/sft/3b_qlora/fsdp_train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_2/sft/3b_qlora/train.yaml b/configs/recipes/llama3_2/sft/3b_qlora/train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_3/sft/70b_full/train.yaml b/configs/recipes/llama3_3/sft/70b_full/train.yaml
@@ -30,7 +30,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_3/sft/70b_lora/train.yaml b/configs/recipes/llama3_3/sft/70b_lora/train.yaml
@@ -29,7 +29,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama3_3/sft/70b_qlora/train.yaml b/configs/recipes/llama3_3/sft/70b_qlora/train.yaml
@@ -30,7 +30,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama4/sft/scout_base_full/train.yaml b/configs/recipes/llama4/sft/scout_base_full/train.yaml
@@ -25,7 +25,6 @@ data:
   train:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama4/sft/scout_instruct_full/train.yaml b/configs/recipes/llama4/sft/scout_instruct_full/train.yaml
@@ -25,7 +25,6 @@ data:
   train:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama4/sft/scout_instruct_lora/train.yaml b/configs/recipes/llama4/sft/scout_instruct_lora/train.yaml
@@ -25,7 +25,6 @@ data:
   train:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/llama4/sft/scout_instruct_qlora/train.yaml b/configs/recipes/llama4/sft/scout_instruct_qlora/train.yaml
@@ -25,7 +25,6 @@ data:
   train:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/phi4/sft/reasoning_plus/full_train.yaml b/configs/recipes/phi4/sft/reasoning_plus/full_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/phi4/sft/reasoning_plus/lora_train.yaml b/configs/recipes/phi4/sft/reasoning_plus/lora_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/phi4/sft/reasoning_plus/qlora_train.yaml b/configs/recipes/phi4/sft/reasoning_plus/qlora_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/qwen3/sft/30b_a3b_lora/train.yaml b/configs/recipes/qwen3/sft/30b_a3b_lora/train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/qwen3/sft/32b_lora/train.yaml b/configs/recipes/qwen3/sft/32b_lora/train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/qwq/sft/full_train.yaml b/configs/recipes/qwq/sft/full_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/qwq/sft/lora_train.yaml b/configs/recipes/qwq/sft/lora_train.yaml
@@ -24,7 +24,6 @@ data:
     datasets:
       - dataset_name: "yahma/alpaca-cleaned" # 51,760 examples
     target_col: "prompt"
-    use_async_dataset: True
 
 training:
   trainer_type: "TRL_SFT"
diff --git a/configs/recipes/qwq/sft/qlora_train.yaml b/configs/recipes/qwq/sft/qlora_train.yaml
