Rename to KinShield-20b
Browse files
README.md
CHANGED
|
@@ -7,7 +7,7 @@ pipeline_tag: text-generation
|
|
| 7 |
tags: [scam-detection, elder-fraud, gpt-oss, lora, qlora, unsloth, dgx-spark]
|
| 8 |
---
|
| 9 |
|
| 10 |
-
# KinShield gpt-oss-20b
|
| 11 |
|
| 12 |
A QLoRA adapter for OpenAI's **gpt-oss-20b** that turns it into [KinShield](https://zx9d4nrkni.execute-api.us-east-1.amazonaws.com/)'s phone-call evidence detector. KinShield is an entry in the AWS "Zero to Shipped" hackathon. The model reads a call transcript and returns strict JSON with:
|
| 13 |
- a risk level (LOW / MEDIUM / HIGH) and a score
|
|
@@ -45,7 +45,7 @@ This model then taught [KinShield-Tiny v3](https://huggingface.co/Solomonwilsonr
|
|
| 45 |
|
| 46 |
```python
|
| 47 |
from unsloth import FastLanguageModel
|
| 48 |
-
model, tok = FastLanguageModel.from_pretrained("Solomonwilsonr/kinshield-
|
| 49 |
max_seq_length=3072, load_in_4bit=True)
|
| 50 |
FastLanguageModel.for_inference(model)
|
| 51 |
SYSTEM = open("system_prompt.txt").read() # KinShield detector prompt, included in this repo
|
|
|
|
| 7 |
tags: [scam-detection, elder-fraud, gpt-oss, lora, qlora, unsloth, dgx-spark]
|
| 8 |
---
|
| 9 |
|
| 10 |
+
# KinShield-20b: a QLoRA fine-tune of gpt-oss-20b
|
| 11 |
|
| 12 |
A QLoRA adapter for OpenAI's **gpt-oss-20b** that turns it into [KinShield](https://zx9d4nrkni.execute-api.us-east-1.amazonaws.com/)'s phone-call evidence detector. KinShield is an entry in the AWS "Zero to Shipped" hackathon. The model reads a call transcript and returns strict JSON with:
|
| 13 |
- a risk level (LOW / MEDIUM / HIGH) and a score
|
|
|
|
| 45 |
|
| 46 |
```python
|
| 47 |
from unsloth import FastLanguageModel
|
| 48 |
+
model, tok = FastLanguageModel.from_pretrained("Solomonwilsonr/kinshield-20b",
|
| 49 |
max_seq_length=3072, load_in_4bit=True)
|
| 50 |
FastLanguageModel.for_inference(model)
|
| 51 |
SYSTEM = open("system_prompt.txt").read() # KinShield detector prompt, included in this repo
|