It was trained on the first 60B tokens of the
Dolma
dataset, so it is merely a research proof-of-concept to test out the methodolgy.
A separate training run was run with the exact same hyperparameters, but using standard fp16 weights.
The comparison can be found in
this wandb report
.
Sample inference code
pip install ai2-olmo
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline, TextStreamer
tokenizer = AutoTokenizer.from_pretrained("NousResearch/OLMo-Bitnet-1B")
model = AutoModelForCausalLM.from_pretrained("NousResearch/OLMo-Bitnet-1B",
torch_dtype=torch.bfloat16, trust_remote_code=True, device_map="auto")
streamer = TextStreamer(tokenizer)
pipe = pipeline("text-generation", model=model, tokenizer=tokenizer, pad_token_id=tokenizer.eos_token_id,
temperature=0.8, repetition_penalty=1.1, do_sample=True,streamer=streamer)
pipe("The capitol of Paris is", max_new_tokens=256)
OLMo-Bitnet-1B huggingface.co is an AI model on huggingface.co that provides OLMo-Bitnet-1B's model effect (), which can be used instantly with this NousResearch OLMo-Bitnet-1B model. huggingface.co supports a free trial of the OLMo-Bitnet-1B model, and also provides paid use of the OLMo-Bitnet-1B. Support call OLMo-Bitnet-1B model through api, including Node.js, Python, http.
OLMo-Bitnet-1B huggingface.co is an online trial and call api platform, which integrates OLMo-Bitnet-1B's modeling effects, including api services, and provides a free online trial of OLMo-Bitnet-1B, you can try OLMo-Bitnet-1B online for free by clicking the link below.
NousResearch OLMo-Bitnet-1B online free url in huggingface.co:
OLMo-Bitnet-1B is an open source model from GitHub that offers a free installation service, and any user can find OLMo-Bitnet-1B on GitHub to install. At the same time, huggingface.co provides the effect of OLMo-Bitnet-1B install, users can directly use OLMo-Bitnet-1B installed effect in huggingface.co for debugging and trial. It also supports api for free installation.