Instructions to use AmPac/trace with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use AmPac/trace with MLX:
# Make sure mlx-lm is installed # pip install --upgrade mlx-lm # if on a CUDA device, also pip install mlx[cuda] # Generate text with mlx-lm from mlx_lm import load, generate model, tokenizer = load("AmPac/trace") prompt = "Once upon a time in" text = generate(model, tokenizer, prompt=prompt, verbose=True) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- MLX LM
How to use AmPac/trace with MLX LM:
Generate or start a chat session
# Install MLX LM uv tool install mlx-lm # Generate some text mlx_lm.generate --model "AmPac/trace" --prompt "Once upon a time"
- Atomic Chat
Real model card. Occupancy 120.131. Load LoRA. Naked 6/12 vs TRACE 12/12.
Browse files
README.md
CHANGED
|
@@ -1,29 +1,62 @@
|
|
| 1 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
|
| 3 |
**Occupancy opens 13 CFR 120.131.**
|
| 4 |
|
| 5 |
-
504/7(a)
|
| 6 |
|
| 7 |
-
|
| 8 |
|
| 9 |
-
|
| 10 |
|
| 11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
```bash
|
| 14 |
python3 -m sba_screen --talk --as-of 2026-01-15 --program 504
|
| 15 |
-
# then: hey can we 504 this apartment they're just gonna rent out, no shop on site
|
| 16 |
-
# then: what if they occupy 100 percent
|
| 17 |
```
|
| 18 |
|
| 19 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
|
| 21 |
-
|
| 22 |
|
| 23 |
-
|
| 24 |
|
| 25 |
-
|
| 26 |
|
| 27 |
## License
|
| 28 |
|
| 29 |
-
`other`. Qwen2.5 (Tongyi Qianwen) + U.S. government SOP/CFR
|
|
|
|
| 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
|
| 16 |
|
| 17 |
**Occupancy opens 13 CFR 120.131.**
|
| 18 |
|
| 19 |
+
LoRA on Qwen2.5-7B-Instruct for 504/7(a) origination screening. Paste a file. It must open the SOP or CFR paragraph in force on the loan-number date. A designated officer still certifies eligibility and credit.
|
| 20 |
|
| 21 |
+
Not an SBA product. Not for applicants. Outputs are drafts. `verified: false`.
|
| 22 |
|
| 23 |
+
Existing buy → [13 CFR 120.131](https://www.ecfr.gov/current/title-13/section-120.131) **(b)** 51%. New construction → **(a)** 60%. Working capital on a 504 → [13 CFR 120.120](https://www.ecfr.gov/current/title-13/section-120.120) **(b)**. Loan number on 2026-10-15 → **SOP 50 10 8.1**.
|
| 24 |
|
| 25 |
+
## Load
|
| 26 |
+
|
| 27 |
+
These files are an MLX LoRA. They are not the staff runtime by themselves.
|
| 28 |
+
|
| 29 |
+
```python
|
| 30 |
+
from mlx_lm import load, generate
|
| 31 |
+
model, tokenizer = load(
|
| 32 |
+
"mlx-community/Qwen2.5-7B-Instruct-4bit",
|
| 33 |
+
adapter_path="AmPac/trace",
|
| 34 |
+
)
|
| 35 |
+
```
|
| 36 |
+
|
| 37 |
+
Staff path (TRACE binds the board, occupancy still 120.131):
|
| 38 |
|
| 39 |
```bash
|
| 40 |
python3 -m sba_screen --talk --as-of 2026-01-15 --program 504
|
|
|
|
|
|
|
| 41 |
```
|
| 42 |
|
| 43 |
+
## ScreenBench (holdout n=12)
|
| 44 |
+
|
| 45 |
+
| Seat | All-pass | False-kill |
|
| 46 |
+
|---|---|---|
|
| 47 |
+
| Naked adapter (this repo) | 6/12 | 0 |
|
| 48 |
+
| Complete system (lake + TRACE bind) | 12/12 | 0 |
|
| 49 |
+
|
| 50 |
+
Do not quote 12/12 as a naked-7B number. Exam-v1 with bind is 76/76, fk 0.
|
| 51 |
+
|
| 52 |
+
## Use
|
| 53 |
|
| 54 |
+
**Intended:** CDFI, CDC, and 7(a) lender staff drafts.
|
| 55 |
|
| 56 |
+
**Not:** applicants, credit decisions, SBA determinations, loading this adapter as a chatbot.
|
| 57 |
|
| 58 |
+
SOP/CFR is **retrieved**. It was not continued-pretraining.
|
| 59 |
|
| 60 |
## License
|
| 61 |
|
| 62 |
+
`other`. Qwen2.5 (Tongyi Qianwen) + U.S. government SOP/CFR + screening code in the AmPac repo. No live borrower files in train.
|