yqsu95 commited on
Commit
eefc94e
·
verified ·
1 Parent(s): e4e605d

Push model using huggingface_hub.

Browse files
Files changed (3) hide show
  1. README.md +6 -12
  2. config.json +5 -5
  3. training_config.json +32 -1
README.md CHANGED
@@ -1,16 +1,10 @@
1
  ---
2
- language: en
3
- license: mit
4
  tags:
5
- - pytorch
6
- - pertTF
7
- - perturbation
8
  ---
9
 
10
- # pertTF perturbation model (tiny, mvc_only)
11
-
12
- This is model files for pertTF (tiny) trained for perturbation prediction using only all losses turned on (5k HVGs)
13
-
14
- ## Usage
15
- ```python
16
- model = HFPerturbationTFModel.from_pretrained('weililab/pert_5k_alltask')
 
1
  ---
 
 
2
  tags:
3
+ - model_hub_mixin
4
+ - pytorch_model_hub_mixin
 
5
  ---
6
 
7
+ This model has been pushed to the Hub using the [PytorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) integration:
8
+ - Code: [More Information Needed]
9
+ - Paper: [More Information Needed]
10
+ - Docs: [More Information Needed]
 
 
 
config.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
  "cell_emb_style": "cls",
3
- "d_hid": null,
4
  "d_model": 32,
5
  "do_dab": false,
6
- "do_mvc": false,
7
  "domain_spec_batchnorm": false,
8
  "dropout": 0.1,
9
  "ecs_threshold": 0.7,
@@ -12,8 +12,8 @@
12
  "input_emb_style": "continuous",
13
  "mvc_decoder_style": "inner product",
14
  "n_bins": 51,
15
- "n_cls": 1,
16
- "n_pert": 1,
17
  "n_ps": 1,
18
  "nhead": 4,
19
  "nlayers": 2,
@@ -29,6 +29,6 @@
29
  "pred_lochness_next": false,
30
  "ps_decoder2_nlayer": 3,
31
  "use_batch_labels": false,
32
- "use_fast_transformer": false,
33
  "vocab": null
34
  }
 
1
  {
2
  "cell_emb_style": "cls",
3
+ "d_hid": 32,
4
  "d_model": 32,
5
  "do_dab": false,
6
+ "do_mvc": true,
7
  "domain_spec_batchnorm": false,
8
  "dropout": 0.1,
9
  "ecs_threshold": 0.7,
 
12
  "input_emb_style": "continuous",
13
  "mvc_decoder_style": "inner product",
14
  "n_bins": 51,
15
+ "n_cls": 19,
16
+ "n_pert": 9854,
17
  "n_ps": 1,
18
  "nhead": 4,
19
  "nlayers": 2,
 
29
  "pred_lochness_next": false,
30
  "ps_decoder2_nlayer": 3,
31
  "use_batch_labels": false,
32
+ "use_fast_transformer": true,
33
  "vocab": null
34
  }
training_config.json CHANGED
@@ -51,5 +51,36 @@
51
  "hvg_col": "highly_variable",
52
  "mvc_masked_train": true,
53
  "max_seq_len": 4001,
54
- "pad_value": -2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
55
  }
 
51
  "hvg_col": "highly_variable",
52
  "mvc_masked_train": true,
53
  "max_seq_len": 4001,
54
+ "n_pert": 9854,
55
+ "nlayers_pert": 4,
56
+ "n_ps": 1,
57
+ "ntoken": 5057,
58
+ "d_model": 32,
59
+ "nhead": 4,
60
+ "d_hid": 32,
61
+ "nlayers": 2,
62
+ "nlayers_cls": 3,
63
+ "n_cls": 19,
64
+ "vocab": null,
65
+ "dropout": 0.1,
66
+ "pad_token": "<pad>",
67
+ "pad_value": -2,
68
+ "do_mvc": true,
69
+ "do_dab": false,
70
+ "use_batch_labels": false,
71
+ "num_batch_labels": null,
72
+ "domain_spec_batchnorm": false,
73
+ "input_emb_style": "continuous",
74
+ "n_bins": 51,
75
+ "cell_emb_style": "cls",
76
+ "mvc_decoder_style": "inner product",
77
+ "ecs_threshold": 0.7,
78
+ "explicit_zero_prob": true,
79
+ "use_fast_transformer": true,
80
+ "fast_transformer_backend": "flash",
81
+ "pre_norm": false,
82
+ "pred_lochness_next": false,
83
+ "ps_decoder2_nlayer": 3,
84
+ "pert_pad_id": null,
85
+ "pert_dim": null
86
  }