Trelis/TinyLlama-1.1B-intermediate-step-480k-1T-chat-llama-style

TinyLlama-1.1B Chat (1 Trillion token checkpoint)

The prompt format is:

f"[INST] {prompt} [INST]"

just like Llama 2 base models.

Note that this model does not emit the end of sequence (< /s >) token well. I am working to update the fine-tuning to improve this.

The model was fine tuned using an adapted filtered Openassistant dataset here.

The base repo follows here:

TinyLlama-1.1B

https://github.com/jzhang38/TinyLlama

The TinyLlama project aims to pretrain a 1.1B Llama model on 3 trillion tokens. With some proper optimization, we can achieve this within a span of "just" 90 days using 16 A100-40G GPUs 🚀🚀. The training has started on 2023-09-01.

We adopted exactly the same architecture and tokenizer as Llama 2. This means TinyLlama can be plugged and played in many open-source projects built upon Llama. Besides, TinyLlama is compact with only 1.1B parameters. This compactness allows it to cater to a multitude of applications demanding a restricted computation and memory footprint.

This Model

This is an intermediate checkpoint with 480K steps and 1007B tokens.

How to use

You will need the transformers>=4.31 Do check the TinyLlama github page for more information.

from transformers import AutoTokenizer
import transformers 
import torch
model = "PY007/TinyLlama-1.1B-intermediate-step-240k-503b"
tokenizer = AutoTokenizer.from_pretrained(model)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

sequences = pipeline(
    'The TinyLlama project aims to pretrain a 1.1B Llama model on 3 trillion tokens. With some proper optimization, we can achieve this within a span of "just" 90 days using 16 A100-40G GPUs 🚀🚀. The training has started on 2023-09-01.',
    do_sample=True,
    top_k=10,
    num_return_sequences=1,
    repetition_penalty=1.5,
    eos_token_id=tokenizer.eos_token_id,
    max_length=500,
)
for seq in sequences:
    print(f"Result: {seq['generated_text']}")