We Keep Coding

AttributeError: module transformers has no attribute TFGPTNeoForCausalLM

I cloned this repository/documentation https://huggingface.co/EleutherAI/gpt-neo-125M
I get the below error whether I run it on google collab or locally. I also installed transformers using this
pip install git+https://github.com/huggingface/transformers
and made sure the configuration file is named as config.json
5 tokenizer = AutoTokenizer.from_pretrained("gpt-neo-125M/",from_tf=True)
----> 6 model = AutoModelForCausalLM.from_pretrained("gpt-neo-125M",from_tf=True)
7
8
3 frames
/usr/local/lib/python3.7/dist-packages/transformers/file_utils.py in __getattr__(self, name)
AttributeError: module transformers has no attribute TFGPTNeoForCausalLM
Full code:
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("EleutherAI/gpt-neo-125M",from_tf=True)
model = AutoModelForCausalLM.from_pretrained("EleutherAI/gpt-neo-125M",from_tf=True)
transformers-cli env results:
transformers version: 4.10.0.dev0
Platform: Linux-4.4.0-19041-Microsoft-x86_64-with-glibc2.29
Python version: 3.8.5
PyTorch version (GPU?): 1.9.0+cpu (False)
Tensorflow version (GPU?): 2.5.0 (False)
Flax version (CPU?/GPU?/TPU?): not installed (NA)
Jax version: not installed
JaxLib version: not installed
Using GPU in script?:
Using distributed or parallel set-up in script?:
Both collab and locally have TensorFlow 2.5.0 version

Try without using from_tf=True flag like below:
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("EleutherAI/gpt-neo-125M")
model = AutoModelForCausalLM.from_pretrained("EleutherAI/gpt-neo-125M")
from_tf expects the pretrained_model_name_or_path (i.e. the first parameter) to be a path to load saved Tensorflow checkpoints from.

My solution was to first edit the source code to remove the line that adds "TF" in front of the package as the correct transformers module is GPTNeoForCausalLM
, but somewhere in the source code it manually added a "TF" in front of it.
Secondly, before cloning the repository it is a must to run
git lfs install.
This link helped me install git lfs properly https://askubuntu.com/questions/799341/how-to-install-git-lfs-on-ubuntu-16-04

Rasterio unable to open .jp2 files

I'm beginning to play with GeoPySpark and am implementing an example notebook.
I successfully retrieved the images
!curl -o /tmp/B01.jp2 http://sentinel-s2-l1c.s3.amazonaws.com/tiles/32/T/NM/2017/1/4/0/B01.jp2
!curl -o /tmp/B09.jp2 http://sentinel-s2-l1c.s3.amazonaws.com/tiles/32/T/NM/2017/1/4/0/B09.jp2
!curl -o /tmp/B10.jp2 http://sentinel-s2-l1c.s3.amazonaws.com/tiles/32/T/NM/2017/1/4/0/B10.jp2
Here is the script:
import rasterio
import geopyspark as gps
import numpy as np
from pyspark import SparkContext
conf = gps.geopyspark_conf(master="local[*]", appName="sentinel-ingest-example")
pysc = SparkContext(conf=conf)
jp2s = ["/tmp/B01.jp2", "/tmp/B09.jp2", "/tmp/B10.jp2"]
arrs = []
for jp2 in jp2s:
with rasterio.open(jp2) as f: #CRASHES HERE
arrs.append(f.read(1))
data = np.array(arrs, dtype=arrs[0].dtype)
data
The script crashes where I placed the marker here, with the following error:
RasterioIOError: '/tmp/B01.jp2' not recognized as a supported file format.
I copy-pasted the example code exactly, ad in the Rasterio docs it even uses .jp2 files in examples.
I'm using the following version of Rasterio, installed with pip3. I do not have Anaconda installed (messes up my Python environments) and do not have GDAL installed (it refuses to, that would be the topic of another question if it is my only solution)
Name: rasterio
Version: 1.1.0
Summary: Fast and direct raster I/O for use with Numpy and SciPy
Home-page: https://github.com/mapbox/rasterio
Author: Sean Gillies
Author-email: sean#mapbox.com
License: BSD
Location: /usr/local/lib/python3.6/dist-packages
Requires: click-plugins, snuggs, numpy, click, attrs, cligj, affine
Required-by:
Why does it refuse to read .jp2 files? Is there maybe a way to convert them to something usable? Or do you know of any example files similar to these ones in an acceptable format?

I was stuck in the same situation.
I used the pyvips package and it's resolved.
import pyvips
image = pyvips.Image.new_from_file("000240.jp2") image.write_to_file("000240.jpg")

Import from Github : How to fix ImportError

I want to use the open source person re-identification library in Python
on Ubuntu 19.04
with Anaconda
no CUDA
in the terminal PyCharm (or not)
Python version 3.7.3
PyTorch version 1.1.0
For that I have to follow instruction like on their deposite git :
git clone https://github.com/Cysu/open-reid.git
cd open-reid
python setup.py install
python examples/softmax_loss.py -d viper -b 64 -j 2 -a resnet50 --logs-dir logs/softmax-loss/viper-resnet50
I receive the following error:
from sklearn.utils.extmath
import pinvh
ImportError: cannot import name 'pinvh'
I have tried to create virtual environments with previous versions of PyTorch (0.4.1, 0.4.0 and 1.0.1) but I always got:
File "examples/softmax_loss.py", line 12, in <module>
from reid import datasets
ModuleNotFoundError: No module named 'reid'
I do not know how to fix it.
EDIT :
Hi thanks for the answer, the problem is that the import are like :
from reid import datasets
from reid import models
from reid.dist_metric import DistanceMetric
from reid.trainers import Trainer
from reid.evaluators import Evaluator
from reid.utils.data import transforms as T
from reid.utils.data.preprocessor import Preprocessor
from reid.utils.logging import Logger
from reid.utils.serialization import load_checkpoint, save_checkpoint
I tried :
from ../reid import datasets
But I got a
File "examples/softmax_loss.py", line 12
from ../reid import datasets
^
SyntaxError: invalid syntax
EDIT 2 :
After re-installing Python 3.7.3 and pytorch 1.1.0 the problem persist with pinvh... I still got this message :
ImportError: cannot import name 'pinvh' from 'sklearn.utils.extmath'
If you can tell me how to fix it or try to tell me if it works please

Since the directory structure is as below:
/(root)-->|
|
|-->reid |--> (contents inside reid)
|
|
|-->examples | -->softmax_loss.py
|
|-->(Other contents in root directory)
It can be observed that reid is not in the same directory as softmax_loss.py, but instead in the parent directory.
So, in the file softmax_loss.py, at line number 12 and below, replace reid with ../reid, this looks for the directory reid in the parent directory.
The other method is to use: import ../reid as R or any other variable; Then use from R import datasets, and so on

utils.extmath.pinvh was deprecated in scikit-learn version 0.19 and removed in version 0.21. The easy fix is therefore to use an earlier version of scikit-learn.

"import ... as .." trigger AttributeError

when I use "import as" like
import tensorflow.python.ops.control_flow_ops as ass
I get "AttributeError: module 'tensorflow' has no attribute 'python'"
but code below like
import tensorflow.python.ops.control_flow_ops
works fine
env: python 3.6.3 + tensorflow 1.3.0

This is an unfortunate combination of TensorFlow's interface sealing (e.g. removing tf.python symbol visibility from import tensorflow as tf) and Python import semantics. The import tensorflow... as syntax will only access public TensorFlow symbols, while from tensorflow... import ... as can also access private symbols with no API stability guarantees.

deepwater with tensorflow on RHEL6

I am trying to get Python + deepwater + tensorflow to run on RHEL 6.7. Using conda, I have installed python 3.6.0, tensorflow 1.1.0 and also gcc 4.8.5. TF is working fine.
I have installed the following libraries using pip install: h2o-3.11.0.3904-py2.py3-none-any.whl and h2o-3.11.0-py2.py3-none-any.whl.
I tried to run the following example from the h2o tutorial
import h2o
from h2o.estimators.deepwater import H2ODeepWaterEstimator
h2o.init()
train = h2o.import_file("https://h2o-public-test-data.s3.amazonaws.com/bigdata/laptop/mnist/train.csv.gz")
features = list(range(0,784))
target = 784
train[target] = train[target].asfactor()
model = H2ODeepWaterEstimator(epochs=100, activation="Rectifier", hidden=[200,200], ignore_const_cols=False,
mini_batch_size=256, input_dropout_ratio=0.1, hidden_dropout_ratios=[0.5,0.5], stopping_rounds=3,
stopping_tolerance=0.05, stopping_metric="misclassification", score_interval=2, score_duty_cycle=0.5,
score_training_samples=1000, score_validation_samples=1000, nfolds=5, gpu=False, seed=1234, backend="tensorflow")
model.train(x=features, y=target, training_frame=train)
The following exception is thrown
Exception: Unable to initialize the native Deep Learning backend: Cannot find TensorFlow native library for OS: linux, architecture: x86_64. See https://github.com/tensorflow/tensorflow/tree/master/java/README.md for possible solutions (such as building the library from source).
Is there anything else that I am missing? Would I need to build the bits from scratch for this platform?