Update README.md
Browse files
README.md
CHANGED
@@ -8,6 +8,22 @@ BLING-QWEN-MINI-TOOL (1.5B)
|
|
8 |
|
9 |
**bling-qwen-mini-tool** is a RAG-finetuned version on Qwen2-1.5B for use in fact-based context question-answering, packaged with 4_K_M GGUF quantization, providing a very fast, very small inference implementation for use on CPUs.
|
10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
To pull the model via API:
|
12 |
|
13 |
from huggingface_hub import snapshot_download
|
|
|
8 |
|
9 |
**bling-qwen-mini-tool** is a RAG-finetuned version on Qwen2-1.5B for use in fact-based context question-answering, packaged with 4_K_M GGUF quantization, providing a very fast, very small inference implementation for use on CPUs.
|
10 |
|
11 |
+
|
12 |
+
## Benchmark Tests
|
13 |
+
|
14 |
+
Evaluated against the benchmark test: RAG-Instruct-Benchmark-Tester
|
15 |
+
1 Test Run with sample=False & temperature=0.0 (deterministic output) - 1 point for correct answer, 0.5 point for partial correct or blank / NF, 0.0 points for incorrect, and -1 points for hallucinations.
|
16 |
+
|
17 |
+
--Accuracy Score: **93.5** correct out of 100
|
18 |
+
--Not Found Classification: 75.0%
|
19 |
+
--Boolean: 87.5%
|
20 |
+
--Math/Logic: 70.0%
|
21 |
+
--Complex Questions (1-5): 3 (Average)
|
22 |
+
--Summarization Quality (1-5): 3 (Average)
|
23 |
+
--Hallucinations: No hallucinations observed in test runs.
|
24 |
+
|
25 |
+
For test run results (and good indicator of target use cases), please see the files ("core_rag_test" and "answer_sheet" in this repo).
|
26 |
+
|
27 |
To pull the model via API:
|
28 |
|
29 |
from huggingface_hub import snapshot_download
|