AnghaBench-armv8-O2-native-clang-full-coder-epoch1-AMD

This model is a fine-tuned version of Qwen/Qwen2.5-Coder-1.5B on the AnghaBench-armv8-O2-native-clang-full_part_00, the AnghaBench-armv8-O2-native-clang-full_part_01, the AnghaBench-armv8-O2-native-clang-full_part_02, the AnghaBench-armv8-O2-native-clang-full_part_03, the AnghaBench-armv8-O2-native-clang-full_part_04, the AnghaBench-armv8-O2-native-clang-full_part_05, the AnghaBench-armv8-O2-native-clang-full_part_06, the AnghaBench-armv8-O2-native-clang-full_part_07, the AnghaBench-armv8-O2-native-clang-full_part_08, the AnghaBench-armv8-O2-native-clang-full_part_09, the AnghaBench-armv8-O2-native-clang-full_part_10, the AnghaBench-armv8-O2-native-clang-full_part_11, the AnghaBench-armv8-O2-native-clang-full_part_12, the AnghaBench-armv8-O2-native-clang-full_part_13, the AnghaBench-armv8-O2-native-clang-full_part_14, the AnghaBench-armv8-O2-native-clang-full_part_15, the AnghaBench-armv8-O2-native-clang-full_part_16, the AnghaBench-armv8-O2-native-clang-full_part_17, the AnghaBench-armv8-O2-native-clang-full_part_18, the AnghaBench-armv8-O2-native-clang-full_part_19, the AnghaBench-armv8-O2-native-clang-full_part_20, the AnghaBench-armv8-O2-native-clang-full_part_21, the AnghaBench-armv8-O2-native-clang-full_part_22, the AnghaBench-armv8-O2-native-clang-full_part_23, the AnghaBench-armv8-O2-native-clang-full_part_24, the AnghaBench-armv8-O2-native-clang-full_part_25, the AnghaBench-armv8-O2-native-clang-full_part_26, the AnghaBench-armv8-O2-native-clang-full_part_27, the AnghaBench-armv8-O2-native-clang-full_part_28, the AnghaBench-armv8-O2-native-clang-full_part_29, the AnghaBench-armv8-O2-native-clang-full_part_30, the AnghaBench-armv8-O2-native-clang-full_part_31, the AnghaBench-armv8-O2-native-clang-full_part_32, the AnghaBench-armv8-O2-native-clang-full_part_33, the AnghaBench-armv8-O2-native-clang-full_part_34, the AnghaBench-armv8-O2-native-clang-full_part_35, the AnghaBench-armv8-O2-native-clang-full_part_36, the AnghaBench-armv8-O2-native-clang-full_part_37, the AnghaBench-armv8-O2-native-clang-full_part_38, the AnghaBench-armv8-O2-native-clang-full_part_39, the AnghaBench-armv8-O2-native-clang-full_part_40, the AnghaBench-armv8-O2-native-clang-full_part_41, the AnghaBench-armv8-O2-native-clang-full_part_42, the AnghaBench-armv8-O2-native-clang-full_part_43, the AnghaBench-armv8-O2-native-clang-full_part_44, the AnghaBench-armv8-O2-native-clang-full_part_45, the AnghaBench-armv8-O2-native-clang-full_part_46, the AnghaBench-armv8-O2-native-clang-full_part_47, the AnghaBench-armv8-O2-native-clang-full_part_48 and the AnghaBench-armv8-O2-native-clang-full_part_49 datasets.

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 512
• total_eval_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1.0

Training results

Framework versions

• Transformers 4.55.0
• Pytorch 2.8.0+rocm6.3
• Datasets 3.6.0
• Tokenizers 0.21.1