shaomai00

why there is some 'tail tokens' after the period？

2

I have trained the code in /Image CoCo ，during the training , I found there always be some 'tail tokens' after the period and some paddings, even the 451th epoch...

what the 'start state s0' exactly is?

1

I saw 'generate a sequence from the start state s0 to maximize its expected end reward ...' in the paper. I'm wondering what the s0 exactly mean? In the code,...

why the Macro F score or Micro F Score are so low in validation.log?

4

I saw this in validation.log Test : Coverage = 719.66, Average Precision = 0.18053248555916795, Micro Precision = 0.06627056672003306, Micro Recall = 0.6256742172322824, Micro F Score = 0.11984691841392547 => Test :...

hello~ I saw your code in scripts/create_test_head_crop_image.py had these params below: scale_choices = [0.75, 1, 1.25] shear_choices = [-0.25, 0, 0.25] rotation_choices = [0, 45, 90, 135, 180, 225, 270,...

Hardware requirements for GLM-chinese-10B

9

I had an 4 * V100 (4*32G) server, but OOM when I tried to finetune the GLM-chinese-10B model. What's the minimal hardware requirements?

BELLE-LLaMA-EXT-7B和BELLE-on-Open-Datasets的问题

5

您好： 测试了一下这两个新的7B模型，发现各自存在一些问题： 1. BELLE-on-Open-Datasets 在中文指令下，会比较高频地乱入一些英文，同样的prompt下BELLE-7B-2M并没有这样的问题； 2. BELLE-LLaMA-EXT-7B模型的指令模版似乎不是"Human: {instruction} \n\nAssistant: "， `prompt = "Human: 写一首中文歌曲，赞美大自然 \n\nAssistant: " input_ids = tokenizer(prompt, return_tensors="pt").input_ids.to(device) generate_ids = model.generate(input_ids, max_new_tokens=300, do_sample = True, top_k = 30,...

如题

在计算output pooling时不用mask吗？

1

if pooling == 'first-last-avg': outputs = [ keras.layers.GlobalAveragePooling1D()(outputs[0]), keras.layers.GlobalAveragePooling1D()(outputs[-1]) ] output = keras.layers.Average()(outputs) elif pooling == 'last-avg': output = keras.layers.GlobalAveragePooling1D()(outputs[-1]) elif pooling == 'cls': output = keras.layers.Lambda(lambda x: x[:, 0])(outputs[-1])...