twscrape-prepared-regression-NeoBERT-3epochs

This model is a fine-tuned version of chandar-lab/NeoBERT on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.5382
• Mse: 0.0003
• Target 0 Mse: 0.0008
• Target 0 Distributions: <wandb.sdk.data_types.image.Image object at 0x7f15100ee7d0>
• Target 0 Error Distribution: <wandb.sdk.data_types.image.Image object at 0x7f14c00f8d30>
• Target 1 Mse: 0.0003
• Target 1 Distributions: <wandb.sdk.data_types.image.Image object at 0x7f14b80baf80>
• Target 1 Error Distribution: <wandb.sdk.data_types.image.Image object at 0x7f148453fdf0>
• Target 2 Mse: 0.0000
• Target 2 Distributions: <wandb.sdk.data_types.image.Image object at 0x7f14845e9ba0>
• Target 2 Error Distribution: <wandb.sdk.data_types.image.Image object at 0x7f148460e3e0>
• Target 3 Mse: 0.0000
• Target 3 Distributions: <wandb.sdk.data_types.image.Image object at 0x7f14845e9e10>
• Target 3 Error Distribution: <wandb.sdk.data_types.image.Image object at 0x7f148439a4a0>

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: 16
• eval_batch_size: 16
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• total_train_batch_size: 256
• total_eval_batch_size: 128
• optimizer: Use adamw_torch 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: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Mse	Target 0 Mse	Target 0 Distributions	Target 0 Error Distribution	Target 1 Mse	Target 1 Distributions	Target 1 Error Distribution	Target 2 Mse	Target 2 Distributions	Target 2 Error Distribution	Target 3 Mse	Target 3 Distributions	Target 3 Error Distribution
1.3816	1.0	1282	1.4598	0.0003	0.0008	<wandb.sdk.data_types.image.Image object at 0x7f1511421c00>	<wandb.sdk.data_types.image.Image object at 0x7f1511422b00>	0.0003	<wandb.sdk.data_types.image.Image object at 0x7f1521bf0910>	<wandb.sdk.data_types.image.Image object at 0x7f1521e65570>	0.0000	<wandb.sdk.data_types.image.Image object at 0x7f1521e2ff70>	<wandb.sdk.data_types.image.Image object at 0x7f151165da20>	0.0000	<wandb.sdk.data_types.image.Image object at 0x7f1521bf32b0>	<wandb.sdk.data_types.image.Image object at 0x7f151165e2c0>
1.5242	2.0	2564	1.4238	0.0003	0.0008	<wandb.sdk.data_types.image.Image object at 0x7f1520347850>	<wandb.sdk.data_types.image.Image object at 0x7f1520125990>	0.0003	<wandb.sdk.data_types.image.Image object at 0x7f14fe3a38b0>	<wandb.sdk.data_types.image.Image object at 0x7f1520347790>	0.0000	<wandb.sdk.data_types.image.Image object at 0x7f1511896ad0>	<wandb.sdk.data_types.image.Image object at 0x7f15117b0c40>	0.0000	<wandb.sdk.data_types.image.Image object at 0x7f14fe445ab0>	<wandb.sdk.data_types.image.Image object at 0x7f14fe677400>
0.7842	3.0	3846	1.5382	0.0003	0.0008	<wandb.sdk.data_types.image.Image object at 0x7f14fe5d4850>	<wandb.sdk.data_types.image.Image object at 0x7f151014aad0>	0.0003	<wandb.sdk.data_types.image.Image object at 0x7f14fe5fc8b0>	<wandb.sdk.data_types.image.Image object at 0x7f15100bfcd0>	0.0000	<wandb.sdk.data_types.image.Image object at 0x7f15100a5bd0>	<wandb.sdk.data_types.image.Image object at 0x7f14fe55a3e0>	0.0000	<wandb.sdk.data_types.image.Image object at 0x7f14fe1ecd90>	<wandb.sdk.data_types.image.Image object at 0x7f14fe296d40>

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.0.1
• Tokenizers 0.21.0