Upload folder using huggingface_hub
Browse files- models/hnet_1stage_L.json +18 -0
- models/hnet_1stage_L.safetensors +3 -0
- models/hnet_1stage_XL.json +18 -0
- models/hnet_2stage_L.json +18 -0
- models/hnet_2stage_L.safetensors +3 -0
- models/hnet_2stage_XL.json +18 -0
- models/hnet_2stage_XL_chinese.json +18 -0
- models/hnet_2stage_XL_code.json +18 -0
models/hnet_1stage_L.json
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"arch_layout": ["m4", ["T22"], "m4"],
|
3 |
+
"d_model": [1024, 1536],
|
4 |
+
"d_intermediate": [0, 4096],
|
5 |
+
"vocab_size": 256,
|
6 |
+
"ssm_cfg": {
|
7 |
+
"chunk_size": 256,
|
8 |
+
"d_conv": 4,
|
9 |
+
"d_state": 128,
|
10 |
+
"expand": 2
|
11 |
+
},
|
12 |
+
"attn_cfg": {
|
13 |
+
"num_heads": [16, 16],
|
14 |
+
"rotary_emb_dim": [32, 48],
|
15 |
+
"window_size": [1023, -1]
|
16 |
+
},
|
17 |
+
"tie_embeddings": false
|
18 |
+
}
|
models/hnet_1stage_L.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:313d9b8fc9adee06e15486862c9a78ddea73945ea03ff3ac3f6f79ebe205484f
|
3 |
+
size 2717648216
|
models/hnet_1stage_XL.json
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"arch_layout": ["m4", ["T24"], "m4"],
|
3 |
+
"d_model": [1024, 2048],
|
4 |
+
"d_intermediate": [0, 5504],
|
5 |
+
"vocab_size": 256,
|
6 |
+
"ssm_cfg": {
|
7 |
+
"chunk_size": 256,
|
8 |
+
"d_conv": 4,
|
9 |
+
"d_state": 128,
|
10 |
+
"expand": 2
|
11 |
+
},
|
12 |
+
"attn_cfg": {
|
13 |
+
"num_heads": [16, 16],
|
14 |
+
"rotary_emb_dim": [32, 64],
|
15 |
+
"window_size": [1023, -1]
|
16 |
+
},
|
17 |
+
"tie_embeddings": false
|
18 |
+
}
|
models/hnet_2stage_L.json
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"arch_layout": ["m4", ["T1m4", ["T26"], "m4T1"], "m4"],
|
3 |
+
"d_model": [1024, 1024, 1536],
|
4 |
+
"d_intermediate": [0, 2816, 4096],
|
5 |
+
"vocab_size": 256,
|
6 |
+
"ssm_cfg": {
|
7 |
+
"chunk_size": 256,
|
8 |
+
"d_conv": 4,
|
9 |
+
"d_state": 128,
|
10 |
+
"expand": 2
|
11 |
+
},
|
12 |
+
"attn_cfg": {
|
13 |
+
"num_heads": [16, 16, 16],
|
14 |
+
"rotary_emb_dim": [32, 32, 48],
|
15 |
+
"window_size": [1023, 1023, -1]
|
16 |
+
},
|
17 |
+
"tie_embeddings": false
|
18 |
+
}
|
models/hnet_2stage_L.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f4a5fa7cad65fcd853b6d185285c0f688a22b8a4c7e97137584415f972c735dc
|
3 |
+
size 3497304728
|
models/hnet_2stage_XL.json
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"arch_layout": ["m4", ["T1m4", ["T27"], "m4T1"], "m4"],
|
3 |
+
"d_model": [1024, 1536, 2048],
|
4 |
+
"d_intermediate": [0, 4096, 5504],
|
5 |
+
"vocab_size": 256,
|
6 |
+
"ssm_cfg": {
|
7 |
+
"chunk_size": 256,
|
8 |
+
"d_conv": 4,
|
9 |
+
"d_state": 128,
|
10 |
+
"expand": 2
|
11 |
+
},
|
12 |
+
"attn_cfg": {
|
13 |
+
"num_heads": [16, 16, 16],
|
14 |
+
"rotary_emb_dim": [32, 48, 64],
|
15 |
+
"window_size": [1023, 1023, -1]
|
16 |
+
},
|
17 |
+
"tie_embeddings": false
|
18 |
+
}
|
models/hnet_2stage_XL_chinese.json
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"arch_layout": ["m4", ["T1m4", ["T30"], "m4T1"], "m4"],
|
3 |
+
"d_model": [1024, 1536, 2048],
|
4 |
+
"d_intermediate": [0, 4096, 5504],
|
5 |
+
"vocab_size": 256,
|
6 |
+
"ssm_cfg": {
|
7 |
+
"chunk_size": 256,
|
8 |
+
"d_conv": 4,
|
9 |
+
"d_state": 128,
|
10 |
+
"expand": 2
|
11 |
+
},
|
12 |
+
"attn_cfg": {
|
13 |
+
"num_heads": [16, 16, 16],
|
14 |
+
"rotary_emb_dim": [32, 48, 64],
|
15 |
+
"window_size": [1023, 1023, -1]
|
16 |
+
},
|
17 |
+
"tie_embeddings": false
|
18 |
+
}
|
models/hnet_2stage_XL_code.json
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"arch_layout": ["m4", ["T1m4", ["T28"], "m4T1"], "m4"],
|
3 |
+
"d_model": [1024, 1536, 2048],
|
4 |
+
"d_intermediate": [0, 4096, 5504],
|
5 |
+
"vocab_size": 256,
|
6 |
+
"ssm_cfg": {
|
7 |
+
"chunk_size": 256,
|
8 |
+
"d_conv": 4,
|
9 |
+
"d_state": 128,
|
10 |
+
"expand": 2
|
11 |
+
},
|
12 |
+
"attn_cfg": {
|
13 |
+
"num_heads": [16, 16, 16],
|
14 |
+
"rotary_emb_dim": [32, 48, 64],
|
15 |
+
"window_size": [1023, 1023, -1]
|
16 |
+
},
|
17 |
+
"tie_embeddings": false
|
18 |
+
}
|