Update pytorch.bin; Upload private eval checkpoint models
Browse files- config.json +38 -0
- mdlm.py +0 -0
- pytorch_model.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"MDLM"
|
| 4 |
+
],
|
| 5 |
+
"attn_backend": "sdpa",
|
| 6 |
+
"auto_map": {
|
| 7 |
+
"AutoConfig": "mdlm.MDLMConfig",
|
| 8 |
+
"AutoModel": "diffusion.MDLM",
|
| 9 |
+
"AutoModelForMaskedLM": "mdlm.MDLM"
|
| 10 |
+
},
|
| 11 |
+
"backbone_config": {
|
| 12 |
+
"_target_": "backbone_dit_legacy.DITLegacy",
|
| 13 |
+
"adaln": true,
|
| 14 |
+
"attn_backend": "sdpa",
|
| 15 |
+
"block_size": null,
|
| 16 |
+
"causal_attention": false,
|
| 17 |
+
"cond_dim": 128,
|
| 18 |
+
"dropout": 0.1,
|
| 19 |
+
"hidden_size": 768,
|
| 20 |
+
"length": 1024,
|
| 21 |
+
"n_heads": 12,
|
| 22 |
+
"num_layers": 12,
|
| 23 |
+
"pretrained_model_name_or_path": null,
|
| 24 |
+
"tie_word_embeddings": true,
|
| 25 |
+
"vocab_size": 50258
|
| 26 |
+
},
|
| 27 |
+
"keep_clean_bos": true,
|
| 28 |
+
"length": 1024,
|
| 29 |
+
"mask_token_id": 50258,
|
| 30 |
+
"model_type": "mdlm",
|
| 31 |
+
"noise_config": null,
|
| 32 |
+
"pad_vocab_size_multiple": null,
|
| 33 |
+
"tokenization_config": null,
|
| 34 |
+
"torch_dtype": "float32",
|
| 35 |
+
"train_on_context": false,
|
| 36 |
+
"transformers_version": "4.52.4",
|
| 37 |
+
"vocab_size": 50258
|
| 38 |
+
}
|
mdlm.py
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1d12283d158431b7b10524c213c06e73d8018ce7008e51a3f150ef540cbe5b7e
|
| 3 |
+
size 678550422
|