Update README.md

README.md

@@ -5,7 +5,7 @@ tags: []
 
 # Model Card for mAInframer-1
 
-Bloob model for auto-complete in the Cobol programming language.
+LLM for the COBOL programming language.
 
 ## Model Details
 
@@ -13,15 +13,12 @@ Bloob model for auto-complete in the Cobol programming language.
 
 mAInframer is a series of models with different parameter counts: 7b, 13b, 34b.
 
-These models are pretrained on code (Codellama base models) and fine-tuned on Cobol code.
+These models are pretrained on code (CodeLlama base models) and fine-tuned on COBOL code.
 
 
-- **Developed by: Bloop** 
-- **Finetuned from model: Codellama** 
-
 ## Uses
 
-This model can be used for code completion and fill-in-the middle (Cobol).
+This model can be used for code completion and fill-in-the middle (COBOL).
 
 ## How to Get Started with the Model
 
@@ -121,7 +118,7 @@ Result:
 
 Follow the format: `<PRE>prefix<SUF>suffix<MID>`
 
-To complete `PROCEDURE DIVISION` and infill `WORKING STORAGE SECTION.` to solve [CobolEval](https://github.com/BloopAI/COBOLEval) problems:
+To complete `PROCEDURE DIVISION` and infill `WORKING STORAGE SECTION.` to solve [COBOLEval](https://github.com/BloopAI/COBOLEval) problems:
 
 ```python
 prompt = '''<PRE>       IDENTIFICATION DIVISION.
@@ -197,16 +194,12 @@ Result:
 
 ## Training Details
 
-Base model: Codellama
+Base model: CodeLlama
 Finetuning type: LoRA
 
-### Training Data
-
-Stack-v2(Cobol) + Stack Overflow (Cobol) + Web scraped Cobol + Synthetic Cobol data
-
-#### Metrics
+### Metrics
 
-[CobolEval](https://github.com/BloopAI/COBOLEval) is an adaptation of HumanEval where the problems are translated to Cobol.
+[COBOLEval](https://github.com/BloopAI/COBOLEval) is an adaptation of HumanEval where the problems are translated to COBOL.
 
 | **Model**            | CobolEval (pass@1) | 
 |----------------------|--------------------|