Add MemOnce checkpoint: config.json
Browse files- config.json +25 -0
config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"NativeMemoryModel"
|
| 4 |
+
],
|
| 5 |
+
"attn_implementation": "sdpa",
|
| 6 |
+
"base_model_name_or_path": "/home/app.t0002931/work/huggingface/hub/Qwen3.5-4B",
|
| 7 |
+
"dtype": "bfloat16",
|
| 8 |
+
"local_files_only": true,
|
| 9 |
+
"lora_alpha": 256,
|
| 10 |
+
"lora_dropout": 0.05,
|
| 11 |
+
"lora_r": 128,
|
| 12 |
+
"memory_views": 1,
|
| 13 |
+
"model_type": "nativememory_exp37",
|
| 14 |
+
"num_memory_tokens": 64,
|
| 15 |
+
"retrieval_embedding_dim": 1024,
|
| 16 |
+
"retrieval_loss_weight": 0.1,
|
| 17 |
+
"retrieval_margin": 2.0,
|
| 18 |
+
"retrieval_score_threshold": 5.0,
|
| 19 |
+
"retrieval_temperature": 0.1,
|
| 20 |
+
"retrieve_blocks": 64,
|
| 21 |
+
"short_state_tokens": 1024,
|
| 22 |
+
"target_len": 4096,
|
| 23 |
+
"transformers_version": "5.14.1",
|
| 24 |
+
"use_cache": false
|
| 25 |
+
}
|