qwen3-0.6b-vericava-posts-v1

This is a model trained from scratch, using parameters of Qwen/Qwen3-0.6B-FP8 on a dataset of my posts on the Internet.

It achieves the following results on the evaluation set:

• Loss: 6.8017

Model description

It generates text resembling what I post on the Internet.

Intended uses & limitations

CAUTION: It may produce something I'd never say.

I do not impose any restriction(s) on the use of this model.

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0005
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 1024
• optimizer: Use OptimizerNames.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_steps: 1000
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss
2.4845	11.1231	100	7.7641
1.7	22.2462	200	6.2579
1.4179	33.3692	300	5.6225
1.2521	44.4923	400	5.4497
1.0905	55.6154	500	5.5389
0.8382	66.7385	600	5.9830
0.5511	77.8615	700	6.3376
0.3364	88.9846	800	6.5791
0.2083	100.0	900	6.8017

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1