Commit ·
ee584e6
1
Parent(s): 8615040
Update DFlash checkpoint for Laguna-S 2.1 1M BF16 (#1)
Browse files- Update DFlash checkpoint for Laguna-S 2.1 1M BF16 (08c233cab782a564bccbf514b225f468ce3dc740)
Co-authored-by: Adam Baranowski <baranowskiadam@users.noreply.huggingface.co>
- config.json +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -10,7 +10,6 @@
|
|
| 10 |
"num_hidden_layers": 6,
|
| 11 |
"num_key_value_heads": 8,
|
| 12 |
"rms_norm_eps": 1e-06,
|
| 13 |
-
"rope_theta": 500000.0,
|
| 14 |
"sliding_window": 512,
|
| 15 |
"vocab_size": 100352,
|
| 16 |
"layer_types": [
|
|
@@ -21,6 +20,7 @@
|
|
| 21 |
"sliding_attention",
|
| 22 |
"sliding_attention"
|
| 23 |
],
|
|
|
|
| 24 |
"gating": "per-head",
|
| 25 |
"architectures": [
|
| 26 |
"DFlashLagunaForCausalLM"
|
|
|
|
| 10 |
"num_hidden_layers": 6,
|
| 11 |
"num_key_value_heads": 8,
|
| 12 |
"rms_norm_eps": 1e-06,
|
|
|
|
| 13 |
"sliding_window": 512,
|
| 14 |
"vocab_size": 100352,
|
| 15 |
"layer_types": [
|
|
|
|
| 20 |
"sliding_attention",
|
| 21 |
"sliding_attention"
|
| 22 |
],
|
| 23 |
+
"rope_theta": 500000.0,
|
| 24 |
"gating": "per-head",
|
| 25 |
"architectures": [
|
| 26 |
"DFlashLagunaForCausalLM"
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2229962896
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9bdfa311a21aa93d5749084d84a25f1930cf948336da6aeaf2f6ef2257439ddf
|
| 3 |
size 2229962896
|