Ali Safaya
commited on
Commit
·
90eac56
1
Parent(s):
79c62fa
initial commit
Browse files- neurocache_adapter/adapter_config.json +38 -0
- neurocache_config.json +27 -0
- neurocache_model.bin +3 -0
neurocache_adapter/adapter_config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "meta-llama/Llama-3.2-3B",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"eva_config": null,
|
| 7 |
+
"exclude_modules": null,
|
| 8 |
+
"fan_in_fan_out": false,
|
| 9 |
+
"inference_mode": false,
|
| 10 |
+
"init_lora_weights": true,
|
| 11 |
+
"layer_replication": null,
|
| 12 |
+
"layers_pattern": null,
|
| 13 |
+
"layers_to_transform": [
|
| 14 |
+
22,
|
| 15 |
+
23,
|
| 16 |
+
24,
|
| 17 |
+
25
|
| 18 |
+
],
|
| 19 |
+
"loftq_config": {},
|
| 20 |
+
"lora_alpha": 128,
|
| 21 |
+
"lora_bias": false,
|
| 22 |
+
"lora_dropout": 0.1,
|
| 23 |
+
"megatron_config": null,
|
| 24 |
+
"megatron_core": "megatron.core",
|
| 25 |
+
"modules_to_save": null,
|
| 26 |
+
"peft_type": "LORA",
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"up_proj",
|
| 32 |
+
"gate_proj",
|
| 33 |
+
"down_proj"
|
| 34 |
+
],
|
| 35 |
+
"task_type": "CAUSAL_LM",
|
| 36 |
+
"use_dora": false,
|
| 37 |
+
"use_rslora": false
|
| 38 |
+
}
|
neurocache_config.json
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"attention_layers": [
|
| 3 |
+
22,
|
| 4 |
+
23,
|
| 5 |
+
24,
|
| 6 |
+
25
|
| 7 |
+
],
|
| 8 |
+
"auto_mapping": null,
|
| 9 |
+
"base_model_name_or_path": "meta-llama/Llama-3.2-3B",
|
| 10 |
+
"cache_dtype": "float16",
|
| 11 |
+
"cache_size": 32768,
|
| 12 |
+
"cache_type": "FIFO",
|
| 13 |
+
"compression_factor": 6,
|
| 14 |
+
"context_size": 1,
|
| 15 |
+
"global_cache": false,
|
| 16 |
+
"inference_mode": true,
|
| 17 |
+
"neighborhood_size": 2,
|
| 18 |
+
"neurocache_type": "ONDEVICE",
|
| 19 |
+
"pooling_window": 8,
|
| 20 |
+
"retrieval_map": {
|
| 21 |
+
"22": 22
|
| 22 |
+
},
|
| 23 |
+
"retrieval_stride": 2,
|
| 24 |
+
"similarity_fn": "l2",
|
| 25 |
+
"task_type": "CAUSAL_LM",
|
| 26 |
+
"topk": 4
|
| 27 |
+
}
|
neurocache_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d5e413cf9f3a138daea509273dc757875197c22b5b8fc2f91cc2eef4b4e19107
|
| 3 |
+
size 721069134
|