metadata

license: apache-2.0
language:
  - hi
tags:
  - text-generation-inference
  - transformers
  - unsloth
  - gemma
  - trl
base_model: unsloth/gemma-7b-bnb-4bit
pipeline_tag: text-generation

Uploaded model

Developed by: Ellight
License: apache-2.0
Finetuned from model : unsloth/gemma-7b-bnb-4bit

This gemma model was trained 2x faster with Unsloth and Huggingface's TRL library.

Hindi-Gemma-2B-instruct (Instruction-tuned)

Hindi-Gemma-2B-instruct is an instruction-tuned Hindi large language model (LLM) with 2 billion parameters, and it is based on Gemma 2B.

TO do inference using the LORA adapters

from unsloth import FastLanguageModel

model, tokenizer = FastLanguageModel.from_pretrained(

model_name = "Ellight/gemma-2b-bnb-4bit", # YOUR MODEL YOU USED FOR TRAINING

max_seq_length = max_seq_length,

dtype = dtype,

load_in_4bit = load_in_4bit,

)

FastLanguageModel.for_inference(model) # Enable native 2x faster inference

prompt = """

Instruction:

{}

Response:

{}"""

inputs = tokenizer( [ prompt.format( "शतरंज बोर्ड पर कितने वर्ग होते हैं?", # instruction "", # output - leave this blank for generation! )

], return_tensors = "pt").to("cuda")

outputs = model.generate(**inputs, max_new_tokens = 64, use_cache = True)

tokenizer.batch_decode(outputs)