zenlm/zen
1
0
Fork 0
mirror of https://github.com/zenlm/zen.git synced 2026-07-26 22:28:05 +00:00
Code Issues Packages Projects Releases Wiki Activity

Correct zen-max training hardware requirements

Browse Source 
- INT4 model is ~370GB (not 60GB)
- Needs 4x A100 80GB or 8x A100 40GB
- Total VRAM ~500GB for QLoRA training
This commit is contained in:
Hanzo Dev
2025-11-17 00:34:47 -08:00
parent 7c5fdae4eb
commit af22b68c0b
1 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
LLM.md
+2 -1
View File
@@ -157,7 +157,8 @@ The Zen AI project embodies clarity through intelligence, providing:
- **Status**: Repository created, configuration files deployed
- **Training Method**:
- QLoRA on HuggingFace Space (no local download needed)
- 4-bit quantized base model (~60GB VRAM)
- INT4 quantized base model (~370GB, 62 shards)
- Requires 4x A100 80GB or 8x A100 40GB GPUs
- LoRA adapters only (~100MB output)
- Cloud-based identity fine-tuning
- **Capabilities**: