Update README.md
Browse files
README.md
CHANGED
|
@@ -6,6 +6,8 @@ parameters and 1,600e18 training FLOPs. Similarly, the checkpoint `mdm-170M-100e
|
|
| 6 |
an MDM model with 170 million non-embedding parameters, 100e18 training FLOPs, and 1% of the dataset subjected
|
| 7 |
to random sequence lengths during pretraining.
|
| 8 |
|
|
|
|
|
|
|
| 9 |
**Conditional generation**: please see the *sharegpt_safetensors* folder.
|
| 10 |
|
| 11 |
**Reverse curse**: please see the *reverse_safetensors* folder
|
|
|
|
| 6 |
an MDM model with 170 million non-embedding parameters, 100e18 training FLOPs, and 1% of the dataset subjected
|
| 7 |
to random sequence lengths during pretraining.
|
| 8 |
|
| 9 |
+
**Math reasoning**: please see the *gsm8k_safetensors* folder.
|
| 10 |
+
|
| 11 |
**Conditional generation**: please see the *sharegpt_safetensors* folder.
|
| 12 |
|
| 13 |
**Reverse curse**: please see the *reverse_safetensors* folder
|