mzwin9870 commited on
Commit
d5efc46
·
verified ·
1 Parent(s): a368008

Trace v9-160 LoRA: occupancy 13 CFR 120.131; adapter 6/12, system 12/12

Browse files
Files changed (3) hide show
  1. README.md +48 -0
  2. adapter_config.json +16 -0
  3. adapters.safetensors +3 -0
README.md ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language: en
3
+ license: other
4
+ library_name: mlx
5
+ base_model: mlx-community/Qwen2.5-7B-Instruct-4bit
6
+ tags:
7
+ - sba
8
+ - "504"
9
+ - "7a"
10
+ - lora
11
+ - mlx
12
+ pipeline_tag: text-generation
13
+ ---
14
+
15
+ # Trace (v9-160)
16
+
17
+ Occupancy opens **13 CFR 120.131**.
18
+
19
+ Open-source 504/7(a) **screening model** for CDFIs and lenders. LoRA on Qwen2.5-7B-Instruct. Not a chatbot. Not a credit engine. Not an SBA product.
20
+
21
+ The unit you run is **adapter + dated lake + TRACE runner**. The adapter file alone is not the product.
22
+
23
+ ```bash
24
+ python3 -m sba_screen --as-of 2026-01-15 --sketch "504 to buy apartments they will rent out and not occupy"
25
+ ```
26
+
27
+ That must open **13 CFR 120.131**.
28
+
29
+ ## ScreenBench (same grader)
30
+
31
+ | Seat | Holdout n=12 | Exam n=76 | False kill |
32
+ |---|---|---|---|
33
+ | This adapter alone | **6/12** | 71/76 | 0 |
34
+ | Complete system (adapter + TRACE) | **12/12** | 76/76 | 0 |
35
+
36
+ Do not quote 12/12 as a naked-7B score.
37
+
38
+ ## Load (MLX)
39
+
40
+ Base: [`mlx-community/Qwen2.5-7B-Instruct-4bit`](https://huggingface.co/mlx-community/Qwen2.5-7B-Instruct-4bit) revision `c26a38f6a37d0a51b4e9a1eb3026530fa35d9fed`.
41
+
42
+ ```text
43
+ adapters.safetensors sha256 a7ef52fe9c4881e3904343dca7f75ee9ba01132d35de26bac424d3470a7f671b
44
+ ```
45
+
46
+ A designated officer still certifies eligibility and credit. Outputs are drafts. Not for applicants.
47
+
48
+ License `other`: Qwen weights (Tongyi Qianwen) + U.S. government SOP/CFR lake + AmPac runner. `verified: false`.
adapter_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "fine_tune_type": "lora",
3
+ "model": "mlx-community/Qwen2.5-7B-Instruct-4bit",
4
+ "lora_parameters": {
5
+ "rank": 16,
6
+ "dropout": 0.0,
7
+ "scale": 20.0
8
+ },
9
+ "num_layers": 16,
10
+ "mask_prompt": true,
11
+ "max_seq_length": 3072,
12
+ "learning_rate": 2e-05,
13
+ "iters": 200,
14
+ "seed": 0,
15
+ "checkpoint": "0000160"
16
+ }
adapters.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7ef52fe9c4881e3904343dca7f75ee9ba01132d35de26bac424d3470a7f671b
3
+ size 92299188