模型微调研发

解决方案

微调研发工作流

需求调研 → 方案评估 → 模型选择 → 数据预处理 → 型训练、数据要素化 → 微调 → 评估 → 优化迭代、应用研发 → 应用上线

立即咨询

模型微调技术方案

Adapter Fusion Tuning

Prefix Tuning

Prompt Tuning

Instruction Tuning

指令微调

SFT, Supervised FineTune

监督微调

PEFT, Parameter-Efficient Fine-Tuning

参数高效的微调方法

PT, P-Tuning v2

xRunda.com

Freeze

监督微调 xRunda.AI

LoRA, Low-Rank Adaptation of Large Language Models

大语言模型的低阶自适应

QLoRa: Quantized LLMs with Low-Rank Adapters

使用低秩适配器的量化大语言模型

Knowledge Distillation

知识蒸馏

RL, Reinforcement learning

强化学习方式微调

Pre-training

预训练

Happy Customers

xrunda was built
for you — it’s a
perfect fit.

xrunda’s goal has been to do whatever is possible to ensure that we have received personalized service as well as our customers.
Chelsea Burnell
Team Leader, 模型微调研发 at Minto Properties
xrunda’s goal has been to do whatever is possible to ensure that we have received personalized service as well as our customers.
Chelsea Burnell
Team Leader, 模型微调研发 at Minto Properties

联系我们

E : guangli@xrunda.com
P : 13810524000

扫描即答
扫描即答