rename 10k -> 4k

Author: yuiseki

recipes/A5000_24GB_x8/i18n-ja-wikipedia-step-10k.yaml recipes/A5000_24GB_x8/i18n-ja-wikipedia-step-4k.yaml

@@ -1,6 +1,6 @@
 target_task: tasks/i18n/ja.md
 base_model_id: TinyLlama/TinyLlama-1.1B-intermediate-step-715k-1.5T
-model_name: TinyLlama-1.5T-ja-wikipedia-step-10k
+model_name: tinyllama-ja-wikipedia-1.5T-step-4k
 output_base_dir: /data/output
 dataset_id: wikimedia/wikipedia
 dataset_load_config: 20231101.ja
@@ -14,7 +14,7 @@ train_claim_gpu_num: 4
 train_per_device_train_batch_size: 8
 train_gradient_accumulation_steps: 4
 train_num_train_epochs: 4
-train_max_steps: 10000
+train_max_steps: 4000
 train_fp16: True
 inference_max_new_tokens: 32
 evaluations:

src/dataset/load.py

@@ -3,8 +3,9 @@
 import yaml
 from datasets.load import load_dataset
 
-load_dataset("oscar")
-load_dataset("cc100")
+# load_dataset("oscar")
+load_dataset("cc100", "en")
+load_dataset("cc100", "ja")
 load_dataset("cerebras/SlimPajama-627B")
 load_dataset("bigcode/starcoderdata")
 load_dataset("Open-Orca/OpenOrca")