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

RLM-Hinglish-to-English-Bidirectional — GGUF (Q8_0)

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:

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.

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

---