OLMo3-190M-zh-full
为零基础 AI 大模型研发训练营（llm001）L04 Full 模型（190M 参数，完整 1 epoch 训练）。
模型配置
- hidden_size: 768, num_layers: 12, num_heads: 12, intermediate_size: 3072
- vocab_size: 48000, sliding_window: 4096
训练配置
- 数据：cmz1024/llm101-olmo3-zh-demo-data
- 训练：H100, num_train_epochs=1.0, bs=16×8=128, lr=5e-4→5e-5, bf16
- 训练集：1,586,532 条 2048-token 序列，约 3.25B train tokens
- 最终指标：train_loss≈3.91, eval_loss≈3.41
用法
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("Nwna/olmo3-190m-zh-full")
tok = AutoTokenizer.from_pretrained("Nwna/olmo3-190m-zh-full")
- Downloads last month
- 26
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for Nwna/olmo3-190m-zh-full
Unable to build the model tree, the base model loops to the model itself. Learn more.