This is the Q8_0 quantized GGUF version of RLM-hinglish-translator-bidir-2, a fully bidirectional Hinglish ↔ English translation model. Converted using Unsloth, this quantized version offers near full-precision quality with significantly reduced memory usage — making it easy to run locally with llama.cpp or Ollama.

What is This Model?

RLM-Hinglish-to-English-Bidirectional translates in both directions between Hinglish (the code-mixed blend of Hindi and English widely used across India) and English — all within a single model. No need for two separate models.

✅ Hinglish → English
✅ English → Hinglish
✅ Chat-format, instruction-following
✅ Optimized for short to medium informal sentences (everyday chat style)

Available File

File	Quantization	Quality	Use Case
`RLM-hinglish-translator-bidir-2.Q8_0.gguf`	Q8_0	Near lossless	Best quality, recommended

Usage

llama.cpp

llama-cli -hf rudrashah/RLM-hinglish-translator-bidir-2-gguf --jinja

Ollama

Run directly:

ollama run rudrashah/RLM-hinglish-translator-bidir-2-gguf

Or create a local model from the GGUF file:

bash

ollama create RLM-hinglish-bidir -f ./Modelfile
ollama run RLM-hinglish-bidir

Python (Transformers)

Simply prefix your text with Hinglish: or English: to set the translation direction.

python

from transformers import TextStreamer

# Hinglish → English
messages = [{
    "role": "user",
    "content": [{"type": "text", "text": "Hinglish: kal mujhe office jaana hai"}]
}]

# English → Hinglish
# messages = [{
#     "role": "user",
#     "content": [{"type": "text", "text": "English: I have to go to the office tomorrow."}]
# }]

inputs = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    return_tensors="pt",
    tokenize=True,
    return_dict=True,
).to("cuda")

_ = model.generate(
    **inputs,
    max_new_tokens=128,
    use_cache=True,
    temperature=1.0, top_p=0.95, top_k=64,
    streamer=TextStreamer(tokenizer, skip_prompt=True),
)

Note: Just change the prefix — use Hinglish: to translate to English, or English: to translate to Hinglish.

Examples

Hinglish → English

Hinglish Input	English Output
`aapka name kya hai?`	`What is your name?`
`kal mujhe office jaana hai`	`I have to go to the office tomorrow.`
`Tumne kal ki match dekhi thi? kon man of the match bana tha?`	`Did you watch yesterday's match? Who was man of the match?`

English → Hinglish

English Input	Hinglish Output
`What is your name?`	`aapka naam kya hai?`
`I have to go to the office tomorrow.`	`kal mujhe office jaana hai.`
`Did you watch yesterday's match?`	`Tumne kal ki match dekhi thi?`

Model Details

Property	Details
Base Model	4B instruction-tuned LLM
Fine-tuning Method	PEFT (LoRA), rank 64, lora_alpha 64
Trainer	TRL SFTTrainer
Dataset	`findnitai/english-to-hinglish`
Training Samples	15,000 (both directions)
Format	Chat template (user/assistant)
Quantization	Q8_0 via Unsloth
License	MIT

Limitations

Best suited for short to medium-length informal sentences, similar to everyday chat messages.
May struggle with heavy regional slang, very long sentences, or dialect-specific expressions.
English → Hinglish outputs reflect the Hinglish style present in the training data and may not match all regional variations.

Model Family

Model	Direction	Format	Notes
RLM-hinglish-translator	Hinglish → English	Full precision	v1, 2B base
RLM-hinglish-translator-2	Hinglish → English	Full precision	v2, 4B base
RLM-hinglish-translator-bidir-2	Hinglish ↔ English	Full precision	Bidirectional
RLM-hinglish-translator-bidir-2-gguf (this)	Hinglish ↔ English	GGUF Q8_0	Quantized for local use