zyfan commited on
Commit
bba1432
·
verified ·
1 Parent(s): 1bd5e47

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -2
README.md CHANGED
@@ -20,7 +20,6 @@ Pre-trained Megatron-format checkpoints for [MUG-V 10B](https://github.com/Shope
20
 
21
  **Torch Distributed Checkpoint** - Flexible parallelism support
22
 
23
- - **Size**: ~64 GB
24
  - **Format**: Torch Distributed (`.distcp`)
25
  - **Parallelism**: Can be loaded with **any TP/PP configuration**
26
  - **Use Case**: Production training, flexible distributed setup
@@ -33,7 +32,6 @@ huggingface-cli download MUG-V/MUG-V-training --local-dir ./checkpoints --includ
33
 
34
  **Torch Format (Legacy)** - Fixed TP=4
35
 
36
- - **Size**: ~64 GB
37
  - **Format**: Torch format (`mp_rank_XX/model_optim_rng.pt`)
38
  - **Parallelism**: Must be loaded with **TP=4**
39
  - **Use Case**: Fixed TP setup or conversion to Torch Distributed
 
20
 
21
  **Torch Distributed Checkpoint** - Flexible parallelism support
22
 
 
23
  - **Format**: Torch Distributed (`.distcp`)
24
  - **Parallelism**: Can be loaded with **any TP/PP configuration**
25
  - **Use Case**: Production training, flexible distributed setup
 
32
 
33
  **Torch Format (Legacy)** - Fixed TP=4
34
 
 
35
  - **Format**: Torch format (`mp_rank_XX/model_optim_rng.pt`)
36
  - **Parallelism**: Must be loaded with **TP=4**
37
  - **Use Case**: Fixed TP setup or conversion to Torch Distributed