#xtuner - Twitter Hashtag

almost 2 years ago

🚀 #XTuner's Zero Memory Waste and Sequence Parallel in Preference Alignment! 🚀 ⚡50% Faster: Cut DPO training time in half with packed preference data. 📈 Sequence Parallel: Train #Llama3 70B RM with sequence lengths up to 1M tokens on 64 A100s. 😉Code: https://t.co/S8I7JMSp9w

intern_lm's tweet photo. 🚀 #XTuner's Zero Memory Waste and Sequence Parallel in Preference Alignment! 🚀
⚡50% Faster: Cut DPO training time in half with packed preference data.
📈 Sequence Parallel: Train #Llama3 70B RM with sequence lengths up to 1M tokens on 64 A100s.
😉Code: https://t.co/S8I7JMSp9w https://t.co/0zikUFqwKL

1

24

7

9

14K

Intern Large Models

@intern_lm

about 2 years ago

🥳#XTuner team releases the new multi-modal model, LLaVA-Llama-3-8B, with #Llama3 LLM, achieving much better performance on various benchmarks. 🥰The performance evaluation substantially surpasses Llama-2. (LLaVA-Llama-3-70B is coming soon!)@AIatMeta 👉https://t.co/S8I7JMSp9w

intern_lm's tweet photo. 🥳#XTuner team releases the new multi-modal model, LLaVA-Llama-3-8B, with #Llama3 LLM, achieving much better performance on various benchmarks.
🥰The performance evaluation substantially surpasses Llama-2. (LLaVA-Llama-3-70B is coming soon!)@AIatMeta
👉https://t.co/S8I7JMSp9w https://t.co/wLV99Vlq8a

3

84

23

37

9K

OpenMMLab @OpenMMLab

over 2 years ago

🥳#XTuner now supports the pre-training and fine-tuning of #LLaVA, achieving much better performance than the official framework (Framework Matters!) 🥰Plus, training costs are further reduced - 7B model only requires 8*48GB + 10 hours 🚀 👉More details: https://t.co/rrWGh1V1PL

OpenMMLab's tweet photo. 🥳#XTuner now supports the pre-training and fine-tuning of #LLaVA, achieving much better performance than the official framework (Framework Matters!)
🥰Plus, training costs are further reduced - 7B model only requires 8*48GB + 10 hours 🚀
👉More details: https://t.co/rrWGh1V1PL https://t.co/WGcWpHgYf7

2

14

1

8

2K

OpenMMLab @OpenMMLab

almost 3 years ago

🥳#XTuner V0.1.2 has been released!#LLMs 🤗Support #Baichuan2 models training and fine-tuning 🤗Add #CodeLlama-Chat conversation template 🤗Redesigned the usage of XTuner convert CLI, for more convenient model conversion 👉https://t.co/rrWGh1Uu0d

OpenMMLab's tweet photo. 🥳#XTuner V0.1.2 has been released!#LLMs
🤗Support #Baichuan2 models training and fine-tuning
🤗Add #CodeLlama-Chat conversation template
🤗Redesigned the usage of XTuner convert CLI, for more convenient model conversion
👉https://t.co/rrWGh1Uu0d https://t.co/15JRT343uM

0

3

0

695

OpenMMLab @OpenMMLab

almost 3 years ago

🥳Feel free to visit https://t.co/0H7mmsdBdx for more information about #XTuner.

OpenMMLab @OpenMMLab

almost 3 years ago

🥳Cool! Use nearly any GPU (even the free resource, e.g., Colab) to fine-tune custom #LLMs！ 🥰Introducing #XTuner, a toolkit for efficiently fine-tuning #LLM, developed by the #MMRazor and #MMDeploy teams. 🤗Feel free to try it out at https://t.co/rrWGh1V1PL

OpenMMLab's tweet photo. 🥳Cool! Use nearly any GPU (even the free resource, e.g., Colab) to fine-tune custom #LLMs！
🥰Introducing #XTuner, a toolkit for efficiently fine-tuning #LLM, developed by the #MMRazor and #MMDeploy teams.
🤗Feel free to try it out at https://t.co/rrWGh1V1PL https://t.co/jpCu7D2wYd

1

39

9

13

4K

0

6

0

1K

OpenMMLab @OpenMMLab

almost 3 years ago

🥳Cool! Use nearly any GPU (even the free resource, e.g., Colab) to fine-tune custom #LLMs！ 🥰Introducing #XTuner, a toolkit for efficiently fine-tuning #LLM, developed by the #MMRazor and #MMDeploy teams. 🤗Feel free to try it out at https://t.co/rrWGh1V1PL

1

39

9

13

4K

IDUTEX Technology @Vpeckertech

almost 9 years ago

#Xtuner T1 update!! ISUZUV12.4 EN 1、Add Euro 5,Euro 4,Euro 3,Euro 2 Series. 2、Add ISUZU Actuations.

1

2

0

IDUTEX Technology @Vpeckertech

almost 9 years ago

#XTUNER T1 update information: RENAULT V12.2 EN 1. Add 35 new modules. 2. Add system scanning and DTC Freeze.

0

1

0

Top Tweets for #xtuner

Last Seen Hashtags on Sotwe

Trends for you

Most Popular Users