Upload config.json with huggingface_hub
Browse files- config.json +12 -11
config.json
CHANGED
|
@@ -1,14 +1,14 @@
|
|
| 1 |
{
|
| 2 |
-
"vocab_size":
|
| 3 |
-
"dim":
|
| 4 |
-
"n_layers":
|
| 5 |
"n_heads": 8,
|
| 6 |
-
"max_seq_len":
|
| 7 |
-
"channel_top_k":
|
| 8 |
-
"token_top_k":
|
| 9 |
"ffn_mult": 4,
|
| 10 |
"mem_dim": 128,
|
| 11 |
-
"mem_size":
|
| 12 |
"mem_k": 8,
|
| 13 |
"mem_threshold": 0.4,
|
| 14 |
"memory_layers": [
|
|
@@ -28,16 +28,17 @@
|
|
| 28 |
"modulation_strength": 0.1,
|
| 29 |
"excitation_strength": 0.3,
|
| 30 |
"dropout": 0.1,
|
| 31 |
-
"batch_size":
|
| 32 |
"lr": 0.0003,
|
| 33 |
-
"max_steps":
|
| 34 |
-
"warmup":
|
| 35 |
"log_every": 100,
|
| 36 |
"eval_every": 500,
|
| 37 |
-
"patience":
|
| 38 |
"use_diversity": true,
|
| 39 |
"pad_id": 0,
|
| 40 |
"eos_id": 2,
|
|
|
|
| 41 |
"architecture": "SparseMind",
|
| 42 |
"model_name": "Cofos Code"
|
| 43 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"vocab_size": 16384,
|
| 3 |
+
"dim": 384,
|
| 4 |
+
"n_layers": 8,
|
| 5 |
"n_heads": 8,
|
| 6 |
+
"max_seq_len": 512,
|
| 7 |
+
"channel_top_k": 128,
|
| 8 |
+
"token_top_k": 96,
|
| 9 |
"ffn_mult": 4,
|
| 10 |
"mem_dim": 128,
|
| 11 |
+
"mem_size": 20000,
|
| 12 |
"mem_k": 8,
|
| 13 |
"mem_threshold": 0.4,
|
| 14 |
"memory_layers": [
|
|
|
|
| 28 |
"modulation_strength": 0.1,
|
| 29 |
"excitation_strength": 0.3,
|
| 30 |
"dropout": 0.1,
|
| 31 |
+
"batch_size": 24,
|
| 32 |
"lr": 0.0003,
|
| 33 |
+
"max_steps": 8000,
|
| 34 |
+
"warmup": 800,
|
| 35 |
"log_every": 100,
|
| 36 |
"eval_every": 500,
|
| 37 |
+
"patience": 8,
|
| 38 |
"use_diversity": true,
|
| 39 |
"pad_id": 0,
|
| 40 |
"eos_id": 2,
|
| 41 |
+
"unk_id": 3,
|
| 42 |
"architecture": "SparseMind",
|
| 43 |
"model_name": "Cofos Code"
|
| 44 |
}
|