Commit
•
fd322cb
1
Parent(s):
ad34de3
End of training
Browse files- README.md +13 -7
- pytorch_model.bin +1 -1
README.md
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
---
|
| 2 |
-
base_model: lordspline/
|
| 3 |
tags:
|
| 4 |
- axolotl
|
| 5 |
- generated_from_trainer
|
|
@@ -16,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
axolotl version: `0.4.1`
|
| 18 |
```yaml
|
| 19 |
-
base_model: lordspline/
|
| 20 |
model_type: AutoModelForCausalLM
|
| 21 |
tokenizer_type: AutoTokenizer
|
| 22 |
|
|
@@ -26,7 +26,13 @@ strict: false
|
|
| 26 |
|
| 27 |
chat_template: chatml
|
| 28 |
datasets:
|
| 29 |
-
- path: lordspline/scidata
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
type: sharegpt
|
| 31 |
conversation: chatml
|
| 32 |
|
|
@@ -64,7 +70,7 @@ gradient_checkpointing: unsloth
|
|
| 64 |
gradient_checkpointing_kwargs:
|
| 65 |
use_reentrant: true # look
|
| 66 |
early_stopping_patience:
|
| 67 |
-
resume_from_checkpoint: ./mergestein/checkpoint-8015
|
| 68 |
local_rank:
|
| 69 |
logging_steps: 1
|
| 70 |
xformers_attention:
|
|
@@ -93,9 +99,9 @@ tokens:
|
|
| 93 |
|
| 94 |
# mergestein
|
| 95 |
|
| 96 |
-
This model is a fine-tuned version of [lordspline/
|
| 97 |
It achieves the following results on the evaluation set:
|
| 98 |
-
- Loss: 1.
|
| 99 |
|
| 100 |
## Model description
|
| 101 |
|
|
@@ -127,7 +133,7 @@ The following hyperparameters were used during training:
|
|
| 127 |
|
| 128 |
| Training Loss | Epoch | Step | Validation Loss |
|
| 129 |
|:-------------:|:-----:|:-----:|:---------------:|
|
| 130 |
-
| 1.
|
| 131 |
|
| 132 |
|
| 133 |
### Framework versions
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model: lordspline/mergestein
|
| 3 |
tags:
|
| 4 |
- axolotl
|
| 5 |
- generated_from_trainer
|
|
|
|
| 16 |
|
| 17 |
axolotl version: `0.4.1`
|
| 18 |
```yaml
|
| 19 |
+
base_model: lordspline/mergestein
|
| 20 |
model_type: AutoModelForCausalLM
|
| 21 |
tokenizer_type: AutoTokenizer
|
| 22 |
|
|
|
|
| 26 |
|
| 27 |
chat_template: chatml
|
| 28 |
datasets:
|
| 29 |
+
# - path: lordspline/scidata
|
| 30 |
+
# type: sharegpt
|
| 31 |
+
# conversation: chatml
|
| 32 |
+
- path: lordspline/wizard_v2_196k_unfiltered
|
| 33 |
+
type: sharegpt
|
| 34 |
+
conversation: chatml
|
| 35 |
+
- path: lordspline/ultrainteract
|
| 36 |
type: sharegpt
|
| 37 |
conversation: chatml
|
| 38 |
|
|
|
|
| 70 |
gradient_checkpointing_kwargs:
|
| 71 |
use_reentrant: true # look
|
| 72 |
early_stopping_patience:
|
| 73 |
+
resume_from_checkpoint: # ./mergestein/checkpoint-8015
|
| 74 |
local_rank:
|
| 75 |
logging_steps: 1
|
| 76 |
xformers_attention:
|
|
|
|
| 99 |
|
| 100 |
# mergestein
|
| 101 |
|
| 102 |
+
This model is a fine-tuned version of [lordspline/mergestein](https://huggingface.co/lordspline/mergestein) on the None dataset.
|
| 103 |
It achieves the following results on the evaluation set:
|
| 104 |
+
- Loss: 1.0348
|
| 105 |
|
| 106 |
## Model description
|
| 107 |
|
|
|
|
| 133 |
|
| 134 |
| Training Loss | Epoch | Step | Validation Loss |
|
| 135 |
|:-------------:|:-----:|:-----:|:---------------:|
|
| 136 |
+
| 1.1202 | 1.0 | 25552 | 1.0348 |
|
| 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 1589947346
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc89906f67ddbeb76efbb37239aa493a9b881a2577994f05671aea509adc0188
|
| 3 |
size 1589947346
|