abertsch72/unlimiformer

Issues

• BookSum_Full BART Baseline script/code

  #66 opened 6 months ago by saxenarohit
  4

• Unable to load dataset

  #64 opened 7 months ago by Ozawa333
  4

• DatasetGenerationError

  #65 opened 7 months ago by pppyb
  1

• LLama2_example output random words

  #60 opened a year ago by KerolosAtef
  1

• Can't run the provided llama2 example

  #59 opened a year ago by KerolosAtef
  6

• Error in running Llama 2 generation example

  #63 opened 10 months ago by OswaldHe
  0

• Hardware Requirement for Running Llama-2 inferences

  #61 opened a year ago by shang-zhu
  2

• How can we use unlimiformer for sequence classification (textual entailment)?

  #62 opened 10 months ago by robinsingh-ai
  0

• Running Unlimiformer with the `forward` method

  #38 opened a year ago by testzer0
  3

• Script utilizing LLM

  #51 opened a year ago by jcgeo9
  1

• GPU VRAM Usage during training

  #58 opened a year ago by KevinD777
  1

• reproducing your results

  #57 opened a year ago by patrickocal
  7

• running unlimiformer inference on multiple gpus

  #29 opened a year ago by kekekawaii2839
  6

• knn_args, unlimiformer_args, tokenizer is not defined

  #32 opened a year ago by laeljh
  1

• support other llms?

  #34 opened a year ago by chaunceyliu30
  3

• Question: During training, the calculation of topk value’s att_weight is different from the classic transformer’s multi-head attention.

  #45 opened a year ago by jjkk123456
  1

• About adding a prefix and input length

  #47 opened a year ago by apapoudakis
  3

• Why is the inference so slow?

  #53 opened a year ago by cckao
  3

• Prompt with Llama-2 stops after "Loading checkpoint shards: 0%"

  #56 opened a year ago by XmasRock
  2

• TypeError: torch_replacement_knn_gpu() got an unexpected keyword argument 'device'

  #25 opened a year ago by jordancole21
  17

• Error Encountered While Running 'run_generation.py' Script

  #48 opened a year ago by arqumk
  1

• Steps to run the code

  #33 opened a year ago by sahulsumra
  5

• IndexError when running inference with Llama-2 model

  #54 opened a year ago by shang-zhu
  3

• Use of other Encode/Decoder Models

  #55 opened a year ago by rdmerillat
  8

• Why "import sled" was commented out in run.py?

  #50 opened a year ago by shi-kejian
  4

• RuntimeError: Expected all tensors to be on the same device, but found at least two devices, ....

  #49 opened a year ago by shi-kejian
  0

• multi-gpu unlimiformer training: Expected all tensors to be on the same device

  #52 opened a year ago by shi-kejian
  4

• Question:too many indices for tensor of dimension 1

  #40 opened a year ago by Lavi11C
  16

• Relative positions in RoPE embeddings

  #46 opened a year ago by AshwinRamachandran2002
  2

• Can unlimiformer work with common fine-tuning methods？

  #23 opened a year ago by mrlzh
  1

• Making Unlimiformer work with decoder models (specifically LLaMA)

  #15 opened a year ago by StrangeTcy
  8

• Why using different calculation methods for the key and value of the cross-attention of the decoder layer in the training and validation stages?

  #44 opened a year ago by jjkk123456
  2

• API server for unlimiformer

  #39 opened a year ago by neubig
  2

• Set max_size to 128 but use 512 tokens

  #43 opened a year ago by adivoj
  2

• Errors on running llama with `test_datastore`

  #41 opened a year ago by wywyWang
  8

• error while training

  #42 opened a year ago by kekekawaii2839
  2

• Unused variable `q_embed` in the Llama's `preprocess_query` method

  #31 opened a year ago by seunghyukoh
  1

• Not really an issue - TrainingArguments are now immutable

  #35 opened a year ago by 9au5a
  2

• Sanity check: VRAM usage on llama-2-7b-chat-hf higher than without Unlimiformer on low tokens?

  #26 opened a year ago by SharkWipf
  6

• About the method `attention_forward_hook`

  #30 opened a year ago by seunghyukoh
  2

• Unable to produce any output with llama 2 summarization example

  #28 opened a year ago by cem2ran
  1

• I Will suggest you simple user interface using gradio.

  #27 opened a year ago by imrankh46
  1

• Error while evaluating

  #20 opened a year ago by MonliH
  2

• Reproduce the +test Unlimiformer setup

  #17 opened a year ago by Leonard907
  7

• Support multilingual model like mt0, mBart ?

  #18 opened a year ago by trannhatquy
  2

• Encoder Only Unlimiformer

  #21 opened a year ago by YHL04
  5

• ImportError: cannot import name 'Unlimiformer' from 'unlimiformer'

  #24 opened a year ago by yungsinatra0
  18

• Working with 8bit and 4bit quantized models

  #19 opened 2 years ago by jordancole21
  10

• Can unlimiformer be trained on mutiple gpus?

  #16 opened 2 years ago by Muxv
  1

• Typing checks fail

  #13 opened 2 years ago by StrangeTcy
  2