Upload config.json with huggingface_hub
Browse files- config.json +43 -0
config.json
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"vocab_size": 8192,
|
| 3 |
+
"dim": 256,
|
| 4 |
+
"n_layers": 6,
|
| 5 |
+
"n_heads": 8,
|
| 6 |
+
"max_seq_len": 384,
|
| 7 |
+
"channel_top_k": 96,
|
| 8 |
+
"token_top_k": 64,
|
| 9 |
+
"ffn_mult": 4,
|
| 10 |
+
"mem_dim": 128,
|
| 11 |
+
"mem_size": 15000,
|
| 12 |
+
"mem_k": 8,
|
| 13 |
+
"mem_threshold": 0.4,
|
| 14 |
+
"memory_layers": [
|
| 15 |
+
2,
|
| 16 |
+
5
|
| 17 |
+
],
|
| 18 |
+
"use_memory": true,
|
| 19 |
+
"n_neuron_types": 7,
|
| 20 |
+
"target_stem_ratio": 0.1,
|
| 21 |
+
"min_stem_ratio": 0.08,
|
| 22 |
+
"stem_plasticity": 0.012,
|
| 23 |
+
"reversion_rate": 0.012,
|
| 24 |
+
"min_age_before_revert": 8,
|
| 25 |
+
"update_interval": 10,
|
| 26 |
+
"baseline_revert_ratio": 0.5,
|
| 27 |
+
"inhibition_strength": 0.08,
|
| 28 |
+
"modulation_strength": 0.1,
|
| 29 |
+
"excitation_strength": 0.3,
|
| 30 |
+
"dropout": 0.1,
|
| 31 |
+
"batch_size": 32,
|
| 32 |
+
"lr": 0.0003,
|
| 33 |
+
"max_steps": 6000,
|
| 34 |
+
"warmup": 500,
|
| 35 |
+
"log_every": 100,
|
| 36 |
+
"eval_every": 500,
|
| 37 |
+
"patience": 6,
|
| 38 |
+
"use_diversity": true,
|
| 39 |
+
"pad_id": 0,
|
| 40 |
+
"eos_id": 2,
|
| 41 |
+
"architecture": "SparseMind",
|
| 42 |
+
"model_name": "Cofos Code"
|
| 43 |
+
}
|