hinglish-song-generator-gguf

Hinglish Song Generator — GGUF (Q4_K_M)

A quantized GGUF version of the fine-tuned Mistral 7B Hinglish Song Generator. Runs locally on CPU or GPU using Ollama, LM Studio, or llama.cpp. No internet required after download.

Give it any context — brand launch, wedding, festival, motivation — and it generates a complete Hinglish song with verse, chorus, and outro.

---

Quick Start

Ollama (Easiest)

# Pull and run directly
ollama run hf.co/Ekansh16/hinglish-song-generator-gguf

# Or create a custom Modelfile for best results

Recommended Modelfile for Ollama:

FROM hf.co/Ekansh16/hinglish-song-generator-gguf

PARAMETER temperature 0.8
PARAMETER top_p 0.92
PARAMETER top_k 50
PARAMETER repeat_penalty 1.2
PARAMETER num_predict 400

SYSTEM "You are an expert Hinglish song lyricist. Write songs mixing English grammar with Hindi/Urdu words naturally. Always use Verse 1, Chorus, Verse 2, Outro structure."

ollama create hinglish-generator -f Modelfile
ollama run hinglish-generator

LM Studio

1. Search Ekansh16/hinglish-song-generator-gguf in the model browser
2. Download hinglish-q4_k_m.gguf
3. Load and start chatting

llama.cpp

./llama-cli \
  -m hinglish-q4_k_m.gguf \
  -p "<s>[INST] A startup called ZapCart launching electric scooters for young Indians on August 1st. Energetic vibe. [/INST]" \
  --temp 0.8 \
  --top-p 0.92 \
  --top-k 50 \
  --repeat-penalty 1.2 \
  -n 400

Python with llama-cpp-python

from llama_cpp import Llama

llm = Llama(
    model_path="hinglish-q4_k_m.gguf",
    n_ctx=1024,
    n_threads=4,
)

context = """
A startup called ZapCart is launching their electric scooter on August 1st.
Target audience is young Indians aged 18-30.
Vibe should be energetic and aspirational.
"""

result = llm(
    f"<s>[INST] {context} [/INST]",
    max_tokens=400,
    temperature=0.8,
    top_p=0.92,
    top_k=50,
    repeat_penalty=1.2,
    stop=["</s>", "[INST]"],
    echo=False,
)

print(result["choices"][0]["text"])

---

Sample Output

Prompt: "A new clothing brand called DesiThreads launching their festive Diwali collection on October 20th for young Indians aged 18-35."

Verse 1:
Lights glow kar rahi hain har gali aur bazaar,
DesiThreads leke aaya naya fashion ka tyohaar.
October 20th ko khulega humne apna darwaaza,
Pura India karenge hum style ke hi baahaar.

Chorus:
Yeh desi vibes, yeh modern swag apna,
Festive season mein pura magic dikha apna.
Dil ki hai yaari aur sapnon ki udan,
Let's celebrate Diwali with DesiThreads ki shaan.

Verse 2:
Traditional designs ka naya twist laaye hain yaar,
Cuts hain trendy aur fit hai ekdum tight,
Zindagi ke safar mein ab look tera bright.

Outro:
Hausla buland hai aur dil mein hai josh,
DesiThreads ke sang aaj manao ye josh.

---

File Details

Q4_K_M retains ~99% of original model quality at 31% of the original size (down from 14.5GB). Recommended for all local inference use cases.

---

Prompt Format

This model uses Mistral instruction format. Always wrap your context in [INST] tags:

<s>[INST] your context here [/INST]

Write rich prompts for best results. Include:

• Brand or event name
• Key dates or details
• Target audience
• Desired mood (fun, emotional, energetic, festive)

Good:

A wedding sangeet ceremony. Groom from Delhi, bride from Jaipur.
Fun, filmy, and emotional Bollywood style.

Too vague:

Make a wedding song

---

Performance

---

Output Style

The model generates Hinglish — English sentence structure with Hindi/Urdu words woven in naturally:

Chal udd chalo yaar, it's time to fly high     ← natural Hinglish
ZapCart pe ride kar, life ho gayi clean        ← brand + Hinglish
Mehnat aur hausla, tera mera ye style          ← Hindi emotional words

Common Hindi/Urdu words used: yaar, dil, zindagi, sapna, apna, mehnat, hausla, manzil, safar, pyaar, dosti, khushi

---

Training Summary

---

Full Precision Version

If you need the full model for further fine-tuning or high-precision inference:

👉 Ekansh16/hinglish-song-generator-merged — 14.5GB, transformers-compatible

---