Update README.md
Browse files
README.md
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
---
|
| 2 |
-
base_model: Qwen/Qwen2.5-
|
| 3 |
library_name: peft
|
| 4 |
license: apache-2.0
|
| 5 |
language:
|
|
@@ -24,7 +24,7 @@ tags:
|
|
| 24 |
|
| 25 |
Russian Contract Extraction LoRA is a domain-specific LoRA adapter for extracting structured information from Russian-language procurement and corporate contracts.
|
| 26 |
|
| 27 |
-
The adapter is fine-tuned on top of **Qwen2.5-
|
| 28 |
|
| 29 |
Unlike generic information extraction models, this adapter focuses on complex real-world Russian contracts, including framework agreements, contracts without fixed amounts, contracts with indirect subject descriptions, advance payment schemes, and various VAT configurations.
|
| 30 |
|
|
@@ -32,7 +32,7 @@ Unlike generic information extraction models, this adapter focuses on complex re
|
|
| 32 |
- **Model type:** LoRA Adapter
|
| 33 |
- **Language(s):** Russian
|
| 34 |
- **License:** Apache-2.0 (inherits base model license)
|
| 35 |
-
- **Finetuned from model:** Qwen/Qwen2.5-
|
| 36 |
|
| 37 |
---
|
| 38 |
|
|
@@ -106,11 +106,11 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
| 106 |
from peft import PeftModel
|
| 107 |
|
| 108 |
base = AutoModelForCausalLM.from_pretrained(
|
| 109 |
-
"Qwen/Qwen2.5-
|
| 110 |
)
|
| 111 |
|
| 112 |
tokenizer = AutoTokenizer.from_pretrained(
|
| 113 |
-
"Qwen/Qwen2.5-
|
| 114 |
)
|
| 115 |
|
| 116 |
model = PeftModel.from_pretrained(
|
|
@@ -168,7 +168,7 @@ Extraction Accuracy
|
|
| 168 |
|
| 169 |
| Model | Accuracy |
|
| 170 |
|--------|----------|
|
| 171 |
-
| Qwen2.5-
|
| 172 |
|
| 173 |
The benchmark evaluates extraction of structured information from Russian procurement contracts.
|
| 174 |
|
|
@@ -252,7 +252,7 @@ Special attention was given to difficult enterprise scenarios:
|
|
| 252 |
|
| 253 |
### Architecture
|
| 254 |
|
| 255 |
-
- Base model: Qwen2.5-
|
| 256 |
- Adapter type: LoRA (PEFT)
|
| 257 |
|
| 258 |
### Output Format
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model: Qwen/Qwen2.5-14B-Instruct
|
| 3 |
library_name: peft
|
| 4 |
license: apache-2.0
|
| 5 |
language:
|
|
|
|
| 24 |
|
| 25 |
Russian Contract Extraction LoRA is a domain-specific LoRA adapter for extracting structured information from Russian-language procurement and corporate contracts.
|
| 26 |
|
| 27 |
+
The adapter is fine-tuned on top of **Qwen2.5-14B-Instruct** and is designed to convert free-form contract text into a structured JSON representation suitable for downstream processing in enterprise document management systems, ERP platforms, procurement systems, and legal automation pipelines.
|
| 28 |
|
| 29 |
Unlike generic information extraction models, this adapter focuses on complex real-world Russian contracts, including framework agreements, contracts without fixed amounts, contracts with indirect subject descriptions, advance payment schemes, and various VAT configurations.
|
| 30 |
|
|
|
|
| 32 |
- **Model type:** LoRA Adapter
|
| 33 |
- **Language(s):** Russian
|
| 34 |
- **License:** Apache-2.0 (inherits base model license)
|
| 35 |
+
- **Finetuned from model:** Qwen/Qwen2.5-14B-Instruct
|
| 36 |
|
| 37 |
---
|
| 38 |
|
|
|
|
| 106 |
from peft import PeftModel
|
| 107 |
|
| 108 |
base = AutoModelForCausalLM.from_pretrained(
|
| 109 |
+
"Qwen/Qwen2.5-14B-Instruct"
|
| 110 |
)
|
| 111 |
|
| 112 |
tokenizer = AutoTokenizer.from_pretrained(
|
| 113 |
+
"Qwen/Qwen2.5-14B-Instruct"
|
| 114 |
)
|
| 115 |
|
| 116 |
model = PeftModel.from_pretrained(
|
|
|
|
| 168 |
|
| 169 |
| Model | Accuracy |
|
| 170 |
|--------|----------|
|
| 171 |
+
| Qwen2.5-14B-Instruct | ~85% |
|
| 172 |
|
| 173 |
The benchmark evaluates extraction of structured information from Russian procurement contracts.
|
| 174 |
|
|
|
|
| 252 |
|
| 253 |
### Architecture
|
| 254 |
|
| 255 |
+
- Base model: Qwen2.5-14B-Instruct
|
| 256 |
- Adapter type: LoRA (PEFT)
|
| 257 |
|
| 258 |
### Output Format
|