Image-Text-to-Text
Transformers
TensorBoard
Safetensors
feature-extraction
conversational
custom_code
xiangan commited on
Commit
746e144
·
verified ·
1 Parent(s): 9206a19

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -1
README.md CHANGED
@@ -31,7 +31,6 @@ Meticulously curated **pre-training and SFT data** with rigorous filtering and q
31
 
32
  - **Ultra-Efficient Training Framework** Complete end-to-end training framework designed for maximum efficiency:
33
  - $16000 total budget for full model training on A100 GPUs ($0.6 per GPU/Hour)
34
- - 45% HFU efficiency in 8k context length
35
  - Built on **MegatronLM** with support for **MoE**, **FP8**, and **long sequence parallelization**
36
  - Optimized codebase for cost-effective scaling
37
 
 
31
 
32
  - **Ultra-Efficient Training Framework** Complete end-to-end training framework designed for maximum efficiency:
33
  - $16000 total budget for full model training on A100 GPUs ($0.6 per GPU/Hour)
 
34
  - Built on **MegatronLM** with support for **MoE**, **FP8**, and **long sequence parallelization**
35
  - Optimized codebase for cost-effective scaling
36