Update README.md
Browse files
README.md
CHANGED
|
@@ -13,8 +13,6 @@ We follow an extremely simple format to organize and manage our models aand data
|
|
| 13 |
|
| 14 |
# Model
|
| 15 |
|
| 16 |
-
### Format
|
| 17 |
-
|
| 18 |
Repo should be named as `midreal/{model_function}_{train_method}_{train_technique}_{base_model}_{date}`
|
| 19 |
|
| 20 |
Model card should include ([example](https://huggingface.co/midreal/passage_sft_qlora_llama3_70b_0704)):
|
|
@@ -45,27 +43,8 @@ Tracking of the training procedure
|
|
| 45 |
Person in charge
|
| 46 |
```
|
| 47 |
|
| 48 |
-
### Usage
|
| 49 |
-
|
| 50 |
-
Models could be uploaded with:
|
| 51 |
-
```
|
| 52 |
-
upload
|
| 53 |
-
```
|
| 54 |
-
|
| 55 |
-
and downloaded with:
|
| 56 |
-
```
|
| 57 |
-
download
|
| 58 |
-
```
|
| 59 |
-
|
| 60 |
-
and updated with:
|
| 61 |
-
```
|
| 62 |
-
update
|
| 63 |
-
```
|
| 64 |
-
|
| 65 |
# Dataset
|
| 66 |
|
| 67 |
-
### Format
|
| 68 |
-
|
| 69 |
Repo should be named as `midreal/{model_function}_{train_method}_{status}_{date}`
|
| 70 |
|
| 71 |
Dataset card should include ([example](https://huggingface.co/datasets/midreal/passage_sft_lmflow_0704)):
|
|
@@ -80,9 +59,7 @@ The schema that the data elements should follow
|
|
| 80 |
Person in charge
|
| 81 |
```
|
| 82 |
|
| 83 |
-
Other information about the dataset
|
| 84 |
-
|
| 85 |
-
### Status
|
| 86 |
|
| 87 |
The production of dataset should somehow follow a pipeline from `raw_data` to `story_data` to `openai` or `lmflow` format.
|
| 88 |
|
|
@@ -94,19 +71,25 @@ The production of dataset should somehow follow a pipeline from `raw_data` to `s
|
|
| 94 |
|
| 95 |
`lmflow` refers to [LMFlow data format](https://optimalscale.github.io/LMFlow/examples/DATASETS.html#data-format).
|
| 96 |
|
| 97 |
-
|
|
|
|
|
|
|
| 98 |
|
| 99 |
-
|
| 100 |
```
|
| 101 |
-
upload
|
| 102 |
```
|
| 103 |
|
| 104 |
-
|
| 105 |
```
|
| 106 |
-
|
|
|
|
| 107 |
```
|
| 108 |
|
| 109 |
-
and
|
| 110 |
```
|
| 111 |
-
|
| 112 |
-
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
|
| 14 |
# Model
|
| 15 |
|
|
|
|
|
|
|
| 16 |
Repo should be named as `midreal/{model_function}_{train_method}_{train_technique}_{base_model}_{date}`
|
| 17 |
|
| 18 |
Model card should include ([example](https://huggingface.co/midreal/passage_sft_qlora_llama3_70b_0704)):
|
|
|
|
| 43 |
Person in charge
|
| 44 |
```
|
| 45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
# Dataset
|
| 47 |
|
|
|
|
|
|
|
| 48 |
Repo should be named as `midreal/{model_function}_{train_method}_{status}_{date}`
|
| 49 |
|
| 50 |
Dataset card should include ([example](https://huggingface.co/datasets/midreal/passage_sft_lmflow_0704)):
|
|
|
|
| 59 |
Person in charge
|
| 60 |
```
|
| 61 |
|
| 62 |
+
Other information about the dataset could also be commented on dataset cards.
|
|
|
|
|
|
|
| 63 |
|
| 64 |
The production of dataset should somehow follow a pipeline from `raw_data` to `story_data` to `openai` or `lmflow` format.
|
| 65 |
|
|
|
|
| 71 |
|
| 72 |
`lmflow` refers to [LMFlow data format](https://optimalscale.github.io/LMFlow/examples/DATASETS.html#data-format).
|
| 73 |
|
| 74 |
+
# Usage
|
| 75 |
+
|
| 76 |
+
We suggest using Huggingface CLI ([docs](https://huggingface.co/docs/huggingface_hub/main/en/guides/cli)).
|
| 77 |
|
| 78 |
+
Once you have installed huggingface-cli and login, models/datasets could be uploaded with:
|
| 79 |
```
|
| 80 |
+
huggingface-cli upload [midreal/repo_id] [local_path] ([path_in_repo]) (--repo-type=dataset)
|
| 81 |
```
|
| 82 |
|
| 83 |
+
e.g.
|
| 84 |
```
|
| 85 |
+
huggingface-cli upload midreal/model ./path/to/curated/data /data/train
|
| 86 |
+
huggingface-cli upload midreal/dataset . . --repo-type=dataset
|
| 87 |
```
|
| 88 |
|
| 89 |
+
and downloaded with:
|
| 90 |
```
|
| 91 |
+
huggingface-cli download midreal/model
|
| 92 |
+
huggingface-cli download midreal/dataset --repo-type dataset
|
| 93 |
+
```
|
| 94 |
+
|
| 95 |
+
If the repo doesn’t exist yet, it will be created automatically.
|