Update README.md

README.md

@@ -3,96 +3,60 @@ language:
 - en
 - hi
 license: apache-2.0
+library_name: transformers
+pipeline_tag: translation
 tags:
 - text-generation-inference
-- transformers
-- unsloth
+- hinglish
+- translation
+- code-switching
 - llama
+- unsloth
 - trl
-- Hinglish
+- lora
+- text-generation-inference
 base_model: unsloth/llama-3-8b-Instruct-bnb-4bit
 datasets:
 - cmu_hinglish_dog
 - suyash2739/Hinglish
 ---
-# Better model
 
-I have just deployed a better model than this on [https://huggingface.co/suyash2739/English_to_Hinglish_fintuned_lamma_3_8b_instruct ]
+# Llama 3 8B — English → Hinglish (CMU Hinglish DoG variant)
 
-# Loss Curve
+An earlier variant of my English → Hinglish translation work: Llama 3 8B Instruct fine-tuned with QLoRA on a [cleaned version](https://huggingface.co/datasets/suyash2739/Hinglish) of the CMU Hinglish DoG conversational dataset.
 
-![image/png](https://cdn-uploads.huggingface.co/production/uploads/65187b234965add2b08b2990/f-qJHUQGxN9yaXym_5u4V.png)
+> **Looking for the recommended model?** The newer variant trained on a curated news-domain corpus produces more fluent Hinglish: [English_to_Hinglish_fintuned_lamma_3_8b_instruct](https://huggingface.co/suyash2739/English_to_Hinglish_fintuned_lamma_3_8b_instruct).
 
-# Evaluation Loss
+## Details
 
-![image/png](https://cdn-uploads.huggingface.co/production/uploads/65187b234965add2b08b2990/6VsNF_rgDjXlubd4x8dMk.png)
+- **Base model:** `unsloth/llama-3-8b-Instruct-bnb-4bit`
+- **Method:** QLoRA (4-bit) with Unsloth + HuggingFace TRL
+- **Training data:** [suyash2739/Hinglish](https://huggingface.co/datasets/suyash2739/Hinglish) — cleaned from [cmu_hinglish_dog](https://huggingface.co/datasets/cmu_hinglish_dog) (conversational domain)
+- **License:** Apache 2.0
 
+## How to use
 
-# Colab Files:
-- Model_Use.ipynb file to use the model
-- Hinglish_train_lamma_3_8b_instruct_.ipynb to see how the model is trained
-
-# Inference:
-
-```
-!pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
-!pip install --no-deps xformers trl peft accelerate bitsandbytes
-```
+Same interface as the main model:
 
 ```python
 from unsloth import FastLanguageModel
-import torch
-max_seq_length = 2048
-dtype = None # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+
-load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False.
 
 model, tokenizer = FastLanguageModel.from_pretrained(
-    model_name = "suyash2739/English_to_Hinglish_cmu_hinglish_dog",
-    max_seq_length = max_seq_length,
-    dtype = dtype,
-    load_in_4bit = load_in_4bit,
+    model_name="suyash2739/English_to_Hinglish_cmu_hinglish_dog",
+    max_seq_length=2048,
+    dtype=None,
+    load_in_4bit=True,
 )
 ```
 
-```python
-prompt = """Translate the input from English to Hinglish to give the response.
-
-### Input:
-{}
-
-### Response:
-{}"""
-
-```
-
-```python
-
-inputs = tokenizer(
-[
-  prompt.format(
-        """This is a fine-tuned Hinglish translation model using Llama 3.""", # input
-        "", # output - leave this blank for generation!
-    )
-], return_tensors = "pt").to("cuda")
-
-from transformers import TextStreamer
-text_streamer = TextStreamer(tokenizer)
-```
-
-```python
-_ = model.generate(**inputs, streamer = text_streamer, max_new_tokens = 2048)
-## ye ek fine-tuned Hinglish translation model hai jisme Llama 3 use kiya gaya hai
-
-```
-
-
+Prompt format: `Translate the input from English to Hinglish to give the response.` followed by `### Input:` and `### Response:` sections.
 
-# Uploaded  model
+## Why two variants?
 
-- **Developed by:** suyash2739
-- **License:** apache-2.0
-- **Finetuned from model :** unsloth/llama-3-8b-Instruct-bnb-4bit
+This model captures conversational, dialogue-style Hinglish (CMU DoG is a document-grounded conversation dataset), while the main model targets news-register Hinglish. Comparing the two illustrates how strongly domain of the parallel corpus shapes code-mixing style in the output.
 
-This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
+## Limitations
 
-[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
+- Conversational-domain training data; formal text may translate awkwardly.
+- Romanized Hinglish only.
+- Inherits base-model and corpus biases.