gpt4all/gpt4all-training/build_map.py

54 lines
1.9 KiB
Python
Raw Normal View History

2023-04-08 15:30:53 -04:00
import numpy as np
from nomic import atlas
import glob
from tqdm import tqdm
from datasets import load_dataset, concatenate_datasets
from sklearn.decomposition import PCA
files = glob.glob("inference/*.jsonl")
print(files)
df = concatenate_datasets([load_dataset("json", data_files=file, split="train") for file in tqdm(files)])
print(len(df))
print(df)
df = df.map(lambda example: {"inputs": [prompt + "\n" + response for prompt, response in zip(example["prompt"], example["response"])]},
batched=True,
num_proc=64)
df = df.map(lambda example: {"trained_on": [int(t) for t in example["is_train"]]},
batched=True,
num_proc=64)
df = df.remove_columns("is_train")
text = df.remove_columns(["labels", "input_ids", "embeddings"])
text_df = [text[i] for i in range(len(text))]
atlas.map_text(text_df, indexed_field="inputs",
2023-04-13 16:58:27 -04:00
name="CHANGE ME!",
2023-04-08 15:30:53 -04:00
colorable_fields=["source", "loss", "trained_on"],
reset_project_if_exists=True,
)
# index is local to train/test split, regenerate
data = df.remove_columns(["labels", "input_ids", "index"])
data = data.add_column("index", list(range(len(data))))
# max embed dim is 2048 for now
# note! this is slow in pyarrow/hf datasets
embeddings = np.array(data["embeddings"])
print("embeddings shape:", embeddings.shape)
embeddings = PCA(n_components=2048).fit_transform(embeddings)
data = data.remove_columns(["embeddings"])
columns = data.to_pandas().to_dict("records")
atlas.map_embeddings(embeddings,
data=columns,
id_field="index",
2023-04-13 16:58:27 -04:00
name="CHANGE ME!",
2023-04-08 15:30:53 -04:00
colorable_fields=["source", "loss", "trained_on"],
2023-04-09 11:12:49 -04:00
build_topic_model=True,
topic_label_field="inputs",
2023-04-08 15:30:53 -04:00
reset_project_if_exists=True,)