sartifyllc/sartify_gemma2-2B-16bit

%%capture
# Installs Unsloth, Xformers (Flash Attention) and all other packages!
!pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
!pip install --no-deps "xformers<0.0.27" "trl<0.9.0" peft accelerate bitsandbytes


from unsloth import FastLanguageModel
import torch
max_seq_length = 2048 # Choose any! We auto support RoPE Scaling internally!
dtype = None # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+
load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False.

model_name = "sartifyllc/sartify_gemma2-2B-16bit"


model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = model_name,
    max_seq_length = max_seq_length,
    dtype = dtype,
    trust_remote_code=True,
    # load_in_4bit = load_in_4bit,
    # token = "hf_...", # use one if using gated models like meta-llama/Llama-2-7b-hf
)

alpaca_prompt = """Hapo chini kuna maelezo ya kazi, pamoja na maelezo ya ziada yanayotoa muktadha zaidi. Andika jibu ambalo linakamilisha ombi hilo ipasavyo.

### Maelezo:
{}

### Ziada:
{}

### Jibu:
{}"""

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

# alpaca_prompt = Copied from above

inputs = tokenizer(
[
    alpaca_prompt.format(
        "Rudia tu kila kitu ninachosema kwa Kiingereza kwa Kiswahili wala usiseme chochote kingine.", # instruction
        "Who is the president of Tanzania?", # input
        "", # output - leave this blank for generation!
    )
], return_tensors = "pt").to("cuda")

from transformers import TextStreamer
text_streamer = TextStreamer(tokenizer)
_ = model.generate(**inputs, streamer = text_streamer, max_new_tokens = 128)