use scale factor 1 and rank 256
Browse files- README.md +8 -5
- pytorch_model.bin +1 -1
README.md
CHANGED
@@ -27,7 +27,7 @@ strict: false
|
|
27 |
datasets:
|
28 |
- path: Vezora/Tested-22k-Python-Alpaca
|
29 |
type: alpaca
|
30 |
-
dataset_prepared_path: prepared-
|
31 |
val_set_size: 0.05
|
32 |
output_dir: ./galore-out
|
33 |
|
@@ -60,6 +60,9 @@ optim_target_modules:
|
|
60 |
- self_attn # for llama
|
61 |
- mlp
|
62 |
|
|
|
|
|
|
|
63 |
|
64 |
train_on_inputs: false
|
65 |
group_by_length: false
|
@@ -93,7 +96,7 @@ special_tokens:
|
|
93 |
|
94 |
This model is a fine-tuned version of [HuggingFaceTB/cosmo-1b](https://huggingface.co/HuggingFaceTB/cosmo-1b) on the None dataset.
|
95 |
It achieves the following results on the evaluation set:
|
96 |
-
- Loss: 0.
|
97 |
|
98 |
## Model description
|
99 |
|
@@ -128,9 +131,9 @@ The following hyperparameters were used during training:
|
|
128 |
| Training Loss | Epoch | Step | Validation Loss |
|
129 |
|:-------------:|:-----:|:----:|:---------------:|
|
130 |
| 0.6299 | 0.0 | 1 | 0.6469 |
|
131 |
-
| 0.
|
132 |
-
| 0.
|
133 |
-
| 0.
|
134 |
|
135 |
|
136 |
### Framework versions
|
|
|
27 |
datasets:
|
28 |
- path: Vezora/Tested-22k-Python-Alpaca
|
29 |
type: alpaca
|
30 |
+
dataset_prepared_path: prepared-qlora
|
31 |
val_set_size: 0.05
|
32 |
output_dir: ./galore-out
|
33 |
|
|
|
60 |
- self_attn # for llama
|
61 |
- mlp
|
62 |
|
63 |
+
optim_args:
|
64 |
+
rank: 256
|
65 |
+
scale: 1
|
66 |
|
67 |
train_on_inputs: false
|
68 |
group_by_length: false
|
|
|
96 |
|
97 |
This model is a fine-tuned version of [HuggingFaceTB/cosmo-1b](https://huggingface.co/HuggingFaceTB/cosmo-1b) on the None dataset.
|
98 |
It achieves the following results on the evaluation set:
|
99 |
+
- Loss: 0.2426
|
100 |
|
101 |
## Model description
|
102 |
|
|
|
131 |
| Training Loss | Epoch | Step | Validation Loss |
|
132 |
|:-------------:|:-----:|:----:|:---------------:|
|
133 |
| 0.6299 | 0.0 | 1 | 0.6469 |
|
134 |
+
| 0.4353 | 0.25 | 215 | 0.4139 |
|
135 |
+
| 0.3721 | 0.5 | 430 | 0.2957 |
|
136 |
+
| 0.3514 | 0.75 | 645 | 0.2426 |
|
137 |
|
138 |
|
139 |
### Framework versions
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3483644214
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1d6d4446971e36016f8a0edcef89d93b65b7256eeb5f29482fa1745a7ebef215
|
3 |
size 3483644214
|