In progress training - batch: 2048
Browse files- config.json +54 -0
config.json
ADDED
|
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"att_dropout": 0.0,
|
| 3 |
+
"att_heads": 16,
|
| 4 |
+
"embed_dim": 1024,
|
| 5 |
+
"ff_dim": 3072,
|
| 6 |
+
"ff_dropout": 0.0,
|
| 7 |
+
"head_dim": 128,
|
| 8 |
+
"kv_heads": 8,
|
| 9 |
+
"layer_types": [
|
| 10 |
+
"stateless",
|
| 11 |
+
"stateful",
|
| 12 |
+
"stateless",
|
| 13 |
+
"stateful",
|
| 14 |
+
"stateless",
|
| 15 |
+
"stateful",
|
| 16 |
+
"stateless",
|
| 17 |
+
"stateful",
|
| 18 |
+
"stateless",
|
| 19 |
+
"stateful",
|
| 20 |
+
"stateless",
|
| 21 |
+
"stateful",
|
| 22 |
+
"stateless",
|
| 23 |
+
"stateful",
|
| 24 |
+
"stateless",
|
| 25 |
+
"stateful",
|
| 26 |
+
"stateless",
|
| 27 |
+
"stateful",
|
| 28 |
+
"stateless",
|
| 29 |
+
"stateful",
|
| 30 |
+
"stateless",
|
| 31 |
+
"stateful",
|
| 32 |
+
"stateless",
|
| 33 |
+
"stateful",
|
| 34 |
+
"stateless",
|
| 35 |
+
"stateful",
|
| 36 |
+
"stateless",
|
| 37 |
+
"stateful"
|
| 38 |
+
],
|
| 39 |
+
"memory_gate_type": "linear",
|
| 40 |
+
"num_layers": 28,
|
| 41 |
+
"padding_idx": 151669,
|
| 42 |
+
"rope_base": 1000000,
|
| 43 |
+
"rope_style": "hf",
|
| 44 |
+
"seq_len": 8192,
|
| 45 |
+
"skip_stm": false,
|
| 46 |
+
"stm_batch_size": 4,
|
| 47 |
+
"stm_size": 4096,
|
| 48 |
+
"tie_embeddings": true,
|
| 49 |
+
"training_cache": false,
|
| 50 |
+
"use_flash_attention": false,
|
| 51 |
+
"use_memory_gate": false,
|
| 52 |
+
"use_separate_memory_projections": true,
|
| 53 |
+
"vocab_size": 151936
|
| 54 |
+
}
|