You can deploy this model and start earning money today!

sanganaka/
qwen3-4B-sa-hi-LORA

Name: qwen3-4B-sa-hi-LORA
Availability: InStock

Qwen/Qwen3-4B-Instruct-2507

finetune

text-generation-inference

endpoints_compatible

Views

No views yet

🔥 Overview

qwen3-sa-hi-sft-lora32 is a Sanskrit → Hindi Machine Translation model fine-tuned on top of Qwen3 (base model) using LoRA (rank 32). The model specializes in translating:

Classical Sanskrit poetry (ślokas, Anuṣṭubh meter)
Epic and Purāṇic Sanskrit
Formal Sanskrit prose into accurate, fluent Hindi prose.

🧠 Model Architecture

Base Model: Qwen3-4B-Instruct-2507
Fine-tuning Method: LoRA
LoRA Rank: 32
LoRA Alpha: r=32, alpha=16 (modify if different)
Framework: LLaMA-Factory / HuggingFace PEFT
Precision: Mixed precision FP16/BF16
Context Length: Inherits from Qwen3 base (<fill exact length>)
Tokenizer: Qwen3 tokenizer (same vocab as base)

📚 Training Data

Dataset used: pretamray/sanskrit_poetry_mt_v3 (or your actual dataset name)

Dataset Composition

Curated Sanskrit → Hindi pairs
Includes:
- Anuṣṭubh epic verse (Ramayana, Mahabharata etc.)
- Traditional ślokas
- Sanskrit prose passages
Format: JSONL
Training Size: 59895
Validation Size: 7500

🔧 Training Configuration

Base Model: Qwen3-4B-instruct-2507
LoRA Rank: 32
LoRA Dropout: 0.05 (if used)
LoRA Target Modules: q_proj, k_proj, v_proj, o_proj
Batch Size: <fill>
Learning Rate: <fill>
Epochs: <fill>
Optimizer: AdamW
Scheduler: cosine
Framework: LLaMA-Factory

💡 Example Usage

python

from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

model = AutoModelForCausalLM.from_pretrained(
    "sanganaka/qwen3-4B-sa-hi-LORA",
    device_map="auto"
)

tokenizer = AutoTokenizer.from_pretrained("pretamray/qwen3-sa-hi-sft-lora32")

text = "कोन्वस्मिन्साम्प्रतं लोके गुणवान्कश्च वीर्यवान्"

inputs = tokenizer(text, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=120)
print(tokenizer.decode(out[0], skip_special_tokens=True))