Upload 2 files
Browse files- config.yaml +44 -0
- model.pt +3 -0
config.yaml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
arrangement: frontloaded
|
| 2 |
+
attn_ratio: 1
|
| 3 |
+
bias: false
|
| 4 |
+
d_ff: 3072
|
| 5 |
+
d_head: 64
|
| 6 |
+
d_model: 768
|
| 7 |
+
dropout: 0.0
|
| 8 |
+
irab_w_delta_init: -0.1
|
| 9 |
+
irab_w_shared_init: 0.05
|
| 10 |
+
layer_types:
|
| 11 |
+
- attn
|
| 12 |
+
- attn
|
| 13 |
+
- attn
|
| 14 |
+
- attn
|
| 15 |
+
- mamba
|
| 16 |
+
- mamba
|
| 17 |
+
- mamba
|
| 18 |
+
- mamba
|
| 19 |
+
- mamba
|
| 20 |
+
- mamba
|
| 21 |
+
- mamba
|
| 22 |
+
- mamba
|
| 23 |
+
mamba_chunk_size: 64
|
| 24 |
+
mamba_d_state: 64
|
| 25 |
+
mamba_expand: 2
|
| 26 |
+
mamba_headdim: 64
|
| 27 |
+
mamba_ngroups: 1
|
| 28 |
+
mamba_ratio: 2
|
| 29 |
+
n_heads: 12
|
| 30 |
+
n_kv_heads: 4
|
| 31 |
+
n_layers: 12
|
| 32 |
+
rms_norm_eps: 1.0e-05
|
| 33 |
+
rope_base: 10000.0
|
| 34 |
+
seq_len: 4096
|
| 35 |
+
sgff_gate_dropout: 0.0
|
| 36 |
+
sgff_max_depth: 16
|
| 37 |
+
tie_embeddings: true
|
| 38 |
+
tope_base_struct: 10.0
|
| 39 |
+
tope_blend_init: 0.0
|
| 40 |
+
tope_max_struct_depth: 64
|
| 41 |
+
use_irab: false
|
| 42 |
+
use_sgff: false
|
| 43 |
+
use_tope: false
|
| 44 |
+
vocab_size: 32768
|
model.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7b3fdf7cb9144c73b4cb6c98760b98c7149d3eecf97bff841adbb43cbef3f154
|
| 3 |
+
size 584347959
|