Downloads · 30 days
244
5% of all-time downloads
RichardErkhov/TinyLlama_-_TinyLlama-1.1B-Chat-v0.3-gguf
TinyLlama_-_TinyLlama-1.1B-Chat-v0.3-gguf is a machine learning model from RichardErkhov. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Downloads · 30 days
244
5% of all-time downloads
All-time downloads
4.9K
Public
Repo size
11.5 GB
Likes
0
Public
Click a slice to open those files.
.gguf13.5 GB · 100%
From the Hugging Face model README
Quantization made by Richard Erkhov.
TinyLlama-1.1B-Chat-v0.3 - GGUF
| Name | Quant method | Size |
|---|---|---|
| TinyLlama-1.1B-Chat-v0.3.Q2_K.gguf | Q2_K | 0.4GB |
| TinyLlama-1.1B-Chat-v0.3.IQ3_XS.gguf | IQ3_XS | 0.44GB |
| TinyLlama-1.1B-Chat-v0.3.IQ3_S.gguf | IQ3_S | 0.47GB |
| TinyLlama-1.1B-Chat-v0.3.Q3_K_S.gguf | Q3_K_S | 0.47GB |
| TinyLlama-1.1B-Chat-v0.3.IQ3_M.gguf | IQ3_M | 0.48GB |
| TinyLlama-1.1B-Chat-v0.3.Q3_K.gguf | Q3_K | 0.51GB |
| TinyLlama-1.1B-Chat-v0.3.Q3_K_M.gguf | Q3_K_M | 0.51GB |
| TinyLlama-1.1B-Chat-v0.3.Q3_K_L.gguf | Q3_K_L | 0.55GB |
| TinyLlama-1.1B-Chat-v0.3.IQ4_XS.gguf | IQ4_XS | 0.57GB |
| TinyLlama-1.1B-Chat-v0.3.Q4_0.gguf | Q4_0 | 0.59GB |
| TinyLlama-1.1B-Chat-v0.3.IQ4_NL.gguf | IQ4_NL | 0.6GB |
| TinyLlama-1.1B-Chat-v0.3.Q4_K_S.gguf | Q4_K_S | 0.6GB |
| TinyLlama-1.1B-Chat-v0.3.Q4_K.gguf | Q4_K | 0.62GB |
| TinyLlama-1.1B-Chat-v0.3.Q4_K_M.gguf | Q4_K_M | 0.62GB |
| TinyLlama-1.1B-Chat-v0.3.Q4_1.gguf | Q4_1 | 0.65GB |
| TinyLlama-1.1B-Chat-v0.3.Q5_0.gguf | Q5_0 | 0.71GB |
| TinyLlama-1.1B-Chat-v0.3.Q5_K_S.gguf | Q5_K_S | 0.71GB |
| TinyLlama-1.1B-Chat-v0.3.Q5_K.gguf | Q5_K | 0.73GB |
| TinyLlama-1.1B-Chat-v0.3.Q5_K_M.gguf | Q5_K_M | 0.73GB |
| TinyLlama-1.1B-Chat-v0.3.Q5_1.gguf | Q5_1 | 0.77GB |
| TinyLlama-1.1B-Chat-v0.3.Q6_K.gguf | Q6_K | 0.84GB |
license: apache-2.0 datasets:
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 is the chat model finetuned on top of PY007/TinyLlama-1.1B-intermediate-step-480k-1T. The dataset used is OpenAssistant/oasst_top1_2023-08-25 following the chatml format.
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-Chat-v0.3"
tokenizer = AutoTokenizer.from_pretrained(model)
pipeline = transformers.pipeline(
"text-generation",
model=model,
torch_dtype=torch.float16,
device_map="auto",
)
CHAT_EOS_TOKEN_ID = 32002
prompt = "How to get in a good university?"
formatted_prompt = (
f"<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant\n"
)
sequences = pipeline(
formatted_prompt,
do_sample=True,
top_k=50,
top_p = 0.9,
num_return_sequences=1,
repetition_penalty=1.1,
max_new_tokens=1024,
eos_token_id=CHAT_EOS_TOKEN_ID,
)
for seq in sequences:
print(f"Result: {seq['generated_text']}")