Update README.md

README.md

@@ -1,22 +1,41 @@
----
-language:
-- en
-license: apache-2.0
-tags:
-- text-generation-inference
-- transformers
-- unsloth
-- gemma
-- trl
-base_model: unsloth/gemma-7b-bnb-4bit
----
-
-# Uploaded  model
-
-- **Developed by:** anudeepadi
-- **License:** apache-2.0
-- **Finetuned from model :** unsloth/gemma-7b-bnb-4bit
-
-This gemma model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
-
-[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
+Telugu Gemma 7B Model
+This is a Telugu language model based on the Gemma 7B model, finetuned on a dataset of Telugu instructions and outputs using the Alpaca format. The model was trained 2x faster using Unsloth and Huggingface's TRL library.
+
+Model Details
+Developed by: Anudeep Adi
+Model architecture: Gemma 7B
+Language: Telugu
+License: Apache 2.0
+Base model: unsloth/gemma-7b-bnb-4bit
+Dataset: telugu_teknium_GPTeacher_general_instruct_filtered_romanized
+Finetuning steps: 60
+Finetuning dataset size: 43,614 examples
+Tags: text-generation-inference, transformers, unsloth, gemma, trl
+Usage
+The model can be used for Telugu language text generation tasks following an instruction and optional input prompt in the Alpaca format:
+
+
+Copy code
+Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
+
+### Instruction:
+{instruction}
+
+### Input:
+{input}
+
+### Response:
+{output}
+Replace {instruction}, {input} and {output} with the relevant text. The model will generate a continuation for the {output} field.
+
+Training Procedure
+The model was finetuned on the telugu_teknium_GPTeacher_general_instruct_filtered_romanized dataset which contains 43,614 examples of Telugu instructions and outputs in the Alpaca format.
+
+Finetuning was performed for 60 steps using Unsloth and the TRL library from Huggingface to enable 2x faster training. LoRA adapters were added to enable efficient finetuning of the 7B parameter model. Mixed precision training with bfloat16 was used.
+
+Unsloth also provides optimized inference code to run the model 2x faster.
+
+<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>
+
+Limitations
+As the model was trained on a relatively small Telugu dataset, its knowledge and conversational abilities in Telugu are limited compared to larger language models. It may make factual errors or inconsistent statements. The model should be used as an experimental prototype only.