Upload config.json with huggingface_hub
Browse files- config.json +11 -29
config.json
CHANGED
|
@@ -1,35 +1,17 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"d_model": 1024,
|
| 4 |
"n_heads": 16,
|
| 5 |
-
"n_kv_heads":
|
| 6 |
-
"d_head":
|
|
|
|
| 7 |
"max_seq_len": 4096,
|
| 8 |
-
"
|
| 9 |
-
"
|
| 10 |
-
"
|
| 11 |
-
"
|
| 12 |
-
"
|
| 13 |
-
"
|
| 14 |
-
"dynmoe_max_k": 4,
|
| 15 |
-
"dynmoe_budget_loss": 0.01,
|
| 16 |
-
"n_recurrent_layers": 3,
|
| 17 |
"t_max": 6,
|
| 18 |
"lora_rank": 32,
|
| 19 |
-
"
|
| 20 |
-
"dropout": 0.1,
|
| 21 |
-
"rope_base": 10000.0,
|
| 22 |
-
"yarn_original_len": 1024,
|
| 23 |
-
"yarn_target_len": 4096,
|
| 24 |
-
"msa_block_size": 32,
|
| 25 |
-
"msa_k_by_level": [
|
| 26 |
-
2,
|
| 27 |
-
4,
|
| 28 |
-
8,
|
| 29 |
-
12,
|
| 30 |
-
16
|
| 31 |
-
],
|
| 32 |
-
"msa_default_k": 8,
|
| 33 |
-
"mtp_depth": 2,
|
| 34 |
-
"mla_latent_dim": 256
|
| 35 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"d_model": 1536,
|
|
|
|
| 3 |
"n_heads": 16,
|
| 4 |
+
"n_kv_heads": 2,
|
| 5 |
+
"d_head": 96,
|
| 6 |
+
"vocab_size": 130569,
|
| 7 |
"max_seq_len": 4096,
|
| 8 |
+
"dropout": 0.1,
|
| 9 |
+
"rope_base": 500000.0,
|
| 10 |
+
"d_ff": 4608,
|
| 11 |
+
"n_trunk_layers": 12,
|
| 12 |
+
"n_recur_layers": 8,
|
| 13 |
+
"n_coda_layers": 4,
|
|
|
|
|
|
|
|
|
|
| 14 |
"t_max": 6,
|
| 15 |
"lora_rank": 32,
|
| 16 |
+
"mtp_depth": 2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
}
|