Upload 2 files
Browse files- config.json +52 -51
- configuration_proteinglm.py +5 -5
config.json
CHANGED
|
@@ -1,52 +1,53 @@
|
|
| 1 |
{
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
}
|
|
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "proteinglm-3b-mlm",
|
| 3 |
+
"add_bias_linear": true,
|
| 4 |
+
"add_qkv_bias": true,
|
| 5 |
+
"apply_query_key_layer_scaling": true,
|
| 6 |
+
"apply_residual_connection_post_layernorm": true,
|
| 7 |
+
"architectures": [
|
| 8 |
+
"ProteinGLMModel"
|
| 9 |
+
],
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"attention_softmax_in_fp32": true,
|
| 12 |
+
"auto_map": {
|
| 13 |
+
"AutoConfig": "configuration_proteinglm.ProteinGLMConfig",
|
| 14 |
+
"AutoModel": "modeling_proteinglm.ProteinGLMForMaskedLM",
|
| 15 |
+
"AutoModelForCausalLM": "modeling_proteinglm.ProteinGLMForCasualLM",
|
| 16 |
+
"AutoModelForMaskedLM": "modeling_proteinglm.ProteinGLMForMaskedLM",
|
| 17 |
+
"AutoModelForSequenceClassification": "modeling_proteinglm.ProteinGLMForSequenceClassification",
|
| 18 |
+
"AutoModelForTokenClassification": "modeling_proteinglm.ProteinGLMForTokenClassification"
|
| 19 |
+
},
|
| 20 |
+
"bias_dropout_fusion": true,
|
| 21 |
+
"deepnorm": true,
|
| 22 |
+
"experts_per_token": 0,
|
| 23 |
+
"ffn_hidden_size": 6832,
|
| 24 |
+
"fp32_residual_connection": false,
|
| 25 |
+
"glu_activation": "geglu",
|
| 26 |
+
"head_num": 1,
|
| 27 |
+
"hidden_dropout": 0.0,
|
| 28 |
+
"hidden_size": 2560,
|
| 29 |
+
"initializer_range": 0.02,
|
| 30 |
+
"is_causal": true,
|
| 31 |
+
"kv_channels": 64,
|
| 32 |
+
"layernorm_epsilon": 1e-05,
|
| 33 |
+
"model_type": "ProteinGLM",
|
| 34 |
+
"moe": false,
|
| 35 |
+
"multi_query_attention": false,
|
| 36 |
+
"multi_query_group_num": 1,
|
| 37 |
+
"num_attention_heads": 40,
|
| 38 |
+
"num_experts": 0,
|
| 39 |
+
"num_layers": 36,
|
| 40 |
+
"padded_vocab_size": 128,
|
| 41 |
+
"post_layer_norm": true,
|
| 42 |
+
"quantization_bit": 0,
|
| 43 |
+
"rmsnorm": false,
|
| 44 |
+
"rotary_embedding_2d": false,
|
| 45 |
+
"seq_length": 1024,
|
| 46 |
+
"torch_dtype": "float32",
|
| 47 |
+
"transformers_version": "4.41.2",
|
| 48 |
+
"untie_head": false,
|
| 49 |
+
"use_cache": true,
|
| 50 |
+
"use_pytorch_sdpa": true,
|
| 51 |
+
"vocab_size": 128
|
| 52 |
+
}
|
| 53 |
+
|
configuration_proteinglm.py
CHANGED
|
@@ -5,12 +5,12 @@ class ProteinGLMConfig(PretrainedConfig):
|
|
| 5 |
model_type = "ProteinGLM"
|
| 6 |
def __init__(
|
| 7 |
self,
|
| 8 |
-
num_layers=
|
| 9 |
padded_vocab_size=128,
|
| 10 |
-
hidden_size=
|
| 11 |
-
ffn_hidden_size=
|
| 12 |
kv_channels=64,
|
| 13 |
-
num_attention_heads=
|
| 14 |
seq_length=1024,
|
| 15 |
hidden_dropout=0.0,
|
| 16 |
attention_dropout=0.0,
|
|
@@ -31,7 +31,7 @@ class ProteinGLMConfig(PretrainedConfig):
|
|
| 31 |
quantization_bit=0,
|
| 32 |
rotary_embedding_2d=False,
|
| 33 |
use_pytorch_sdpa=True,
|
| 34 |
-
is_causal=
|
| 35 |
use_cache=True,
|
| 36 |
initializer_range=0.02,
|
| 37 |
moe=False,
|
|
|
|
| 5 |
model_type = "ProteinGLM"
|
| 6 |
def __init__(
|
| 7 |
self,
|
| 8 |
+
num_layers=36,
|
| 9 |
padded_vocab_size=128,
|
| 10 |
+
hidden_size=2560,
|
| 11 |
+
ffn_hidden_size=6832,
|
| 12 |
kv_channels=64,
|
| 13 |
+
num_attention_heads=40,
|
| 14 |
seq_length=1024,
|
| 15 |
hidden_dropout=0.0,
|
| 16 |
attention_dropout=0.0,
|
|
|
|
| 31 |
quantization_bit=0,
|
| 32 |
rotary_embedding_2d=False,
|
| 33 |
use_pytorch_sdpa=True,
|
| 34 |
+
is_causal=True,
|
| 35 |
use_cache=True,
|
| 36 |
initializer_range=0.02,
|
| 37 |
moe=False,
|