ディレクトリの違いだけでディープラーニングの結果に差がある

まったく同じコード、同じデータで機械学習させた結果以下のように学習の仕方が異なります。
lossは-1に近づくほど損失が小さいようにlogで表現しています。
異なっている条件はうまく学習している上のコードがjupyternotebookで作成したディレクトリ、下のval_lossが早くに収束しているほうがjupyternotebook上のpc上のデスクトップのディレクトリにある点だけです。
1度だけではなく必ず起こります。原因が不明なのでわかる方ご教授お願いいたします。
また、同じデータセットを使った場合に以前ならうまく学習していたのに、今日やってみると以前のようには学習しませんでした。結果に差がありすぎて異変を感じました。これもなぜかわかる方いらっしゃいましたらご教授お願いします。
gpuがかなり熱をもって熱くなっていることが以前とは違っていますが、下の結果は同じ日に連続で取得した結果ですので、gpuの問題ではないような気もします。
コードは全く同じで、入れているディレクトリの違いだけなので、コードは載せていません。

環境
jupyter notebook
python3.7
gpuでディープラーニングを行っています。

Train on 800 samples, validate on 200 samples
Epoch 1/30
800/800 [==============================] - 14s 17ms/step - loss: -0.7091 - val_loss: -0.8009
Epoch 2/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8573 - val_loss: -0.8809
Epoch 3/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9014 - val_loss: -0.9143
Epoch 4/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9217 - val_loss: -0.9253
Epoch 5/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9374 - val_loss: -0.9341
Epoch 6/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9469 - val_loss: -0.9449
Epoch 7/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9561 - val_loss: -0.9554
Epoch 8/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9590 - val_loss: -0.9552
Epoch 9/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9642 - val_loss: -0.9611
Epoch 10/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9672 - val_loss: -0.9590
Epoch 11/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9693 - val_loss: -0.9646
Epoch 12/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9718 - val_loss: -0.9618
Epoch 13/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9731 - val_loss: -0.9654
Epoch 14/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9753 - val_loss: -0.9674
Epoch 15/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9769 - val_loss: -0.9677
Epoch 16/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9780 - val_loss: -0.9680
Epoch 17/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9791 - val_loss: -0.9688
Epoch 18/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9798 - val_loss: -0.9691
Epoch 19/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9814 - val_loss: -0.9684
Epoch 20/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9823 - val_loss: -0.9707
Epoch 21/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9831 - val_loss: -0.9705
Epoch 22/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9834 - val_loss: -0.9707
Epoch 23/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9845 - val_loss: -0.9710
Epoch 24/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9848 - val_loss: -0.9690
Epoch 25/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9852 - val_loss: -0.9708
Epoch 26/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9857 - val_loss: -0.9714
Epoch 27/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9858 - val_loss: -0.9720
Epoch 28/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9868 - val_loss: -0.9717
Epoch 29/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9870 - val_loss: -0.9724
Epoch 30/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9873 - val_loss: -0.9723

Train on 800 samples, validate on 200 samples
Epoch 1/30
800/800 [==============================] - 13s 17ms/step - loss: -0.5863 - val_loss: -0.6294
Epoch 2/30
800/800 [==============================] - 7s 9ms/step - loss: -0.6429 - val_loss: -0.6434
Epoch 3/30
800/800 [==============================] - 7s 9ms/step - loss: -0.6645 - val_loss: -0.6661
Epoch 4/30
800/800 [==============================] - 7s 9ms/step - loss: -0.6769 - val_loss: -0.6746
Epoch 5/30
800/800 [==============================] - 7s 9ms/step - loss: -0.6918 - val_loss: -0.6900
Epoch 6/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7017 - val_loss: -0.6966
Epoch 7/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7112 - val_loss: -0.7051
Epoch 8/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7211 - val_loss: -0.7065
Epoch 9/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7352 - val_loss: -0.7185
Epoch 10/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7423 - val_loss: -0.7099
Epoch 11/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7495 - val_loss: -0.7130
Epoch 12/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7573 - val_loss: -0.7185
Epoch 13/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7698 - val_loss: -0.7166
Epoch 14/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7802 - val_loss: -0.7116
Epoch 15/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7876 - val_loss: -0.7063
Epoch 16/30
800/800 [==============================] - 7s 9ms/step - loss: -0.7996 - val_loss: -0.7012
Epoch 17/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8116 - val_loss: -0.7056
Epoch 18/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8213 - val_loss: -0.6921
Epoch 19/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8334 - val_loss: -0.6884
Epoch 20/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8475 - val_loss: -0.6806
Epoch 21/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8582 - val_loss: -0.6839
Epoch 22/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8638 - val_loss: -0.6741
Epoch 23/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8764 - val_loss: -0.6785
Epoch 24/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8842 - val_loss: -0.6700
Epoch 25/30
800/800 [==============================] - 7s 9ms/step - loss: -0.8966 - val_loss: -0.6636
Epoch 26/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9003 - val_loss: -0.6650
Epoch 27/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9078 - val_loss: -0.6632
Epoch 28/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9130 - val_loss: -0.6644
Epoch 29/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9189 - val_loss: -0.6613
Epoch 30/30
800/800 [==============================] - 7s 9ms/step - loss: -0.9259 - val_loss: -0.6661