Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / llm-attacks/llm-attacks issues and pull requests
#108 - Attacking newer models such as Llama-3+
Issue -
State: open - Opened by simsal0r 10 days ago
#107 - Request for categories of harmful strings
Issue -
State: open - Opened by xpq-tech 16 days ago
#106 - Code Doesn't Run Out of The Box ? [Bug Reports]
Issue -
State: open - Opened by hammoudhasan about 2 months ago
#105 - Unstable loss curve in `demo.ipynb`
Issue -
State: closed - Opened by XitaoLi 3 months ago
- 7 comments
#103 - [BUG] self.conv_template.append_message(self.conv_template.roles[0], None)
Issue -
State: open - Opened by Anson-He 3 months ago
- 7 comments
#102 - Error Dataset
Issue -
State: open - Opened by jiah-li 3 months ago
- 2 comments
#101 - Models
Issue -
State: open - Opened by wli565 3 months ago
- 1 comment
#100 - Update attack_manager.py, repaired a bug about the slice of llama2
Pull Request -
State: open - Opened by leileqiTHU 4 months ago
#99 - Problem about the input slice of Llama2-chat-7b-hf
Issue -
State: open - Opened by leileqiTHU 4 months ago
- 1 comment
#98 - How to define local model paths? Use `repo_type` argument if needed.
Issue -
State: closed - Opened by windowshopr 4 months ago
- 2 comments
#97 - Using Multiple GPUs getting multiprocessing error (2 GPUs Each 24 GB)
Issue -
State: open - Opened by NITESHCHACHRIYA 4 months ago
- 6 comments
#96 - Can't Reproduce the paper output
Issue -
State: open - Opened by MariamMatar 4 months ago
- 3 comments
#95 - Control the attack suffix length
Issue -
State: closed - Opened by SinHanYang 5 months ago
- 2 comments
#94 - Out of Memory Error
Issue -
State: open - Opened by samvikum 5 months ago
- 2 comments
#93 - Completed Project Bhagirathi - Redlist Loss GCG for Efficient Attacks Against Aligned LLMs
Pull Request -
State: closed - Opened by HarshavardhanK 6 months ago
- 2 comments
#92 - Open source LLMs
Pull Request -
State: closed - Opened by yasu52 7 months ago
#91 - Cleanup
Pull Request -
State: closed - Opened by vermouthdky 7 months ago
#90 - multiple gpus
Issue -
State: open - Opened by ShiJiawenwen 7 months ago
- 2 comments
#89 - Goal-target Mismatch in AdvBench Harmful Behaviors
Issue -
State: open - Opened by fly-dust 8 months ago
#88 - Update requirements.txt
Pull Request -
State: open - Opened by sihatafnan 8 months ago
#87 - Update requirements.txt
Pull Request -
State: closed - Opened by sihatafnan 8 months ago
#86 - Conversation template of `llama-2` does not work with empty instructions
Issue -
State: open - Opened by nurlanov-zh 8 months ago
- 1 comment
#85 - RuntimeError: CUDA driver error: invalid argument
Issue -
State: open - Opened by roywang021 9 months ago
#84 - How to replicate the experiment using multiple GPUs
Issue -
State: closed - Opened by Laz4rz 10 months ago
- 4 comments
#83 - How to use the learned logfile to evaluate under individual settings
Issue -
State: open - Opened by 4hebailanc 10 months ago
#82 - What are "id_id", "id_od", "od_id", "od_od"?
Issue -
State: open - Opened by shizhouxing 10 months ago
- 2 comments
#81 - The size of tensor a (20) must match the size of tensor b (21) at non-singleton dimension 0
Issue -
State: open - Opened by xukefaker 10 months ago
- 1 comment
#80 - Why is the output result of the demo code looks like the repeated prompt?
Issue -
State: closed - Opened by Dilemma111 10 months ago
#79 - NotImplementedError: Cannot copy out of meta tensor; no data!
Issue -
State: open - Opened by zhu111-eng 11 months ago
#78 - Ambiguity about the batch-size parameter
Issue -
State: open - Opened by Framartin 11 months ago
- 3 comments
#77 - Disable weights gradient
Pull Request -
State: open - Opened by Framartin 11 months ago
- 1 comment
#76 - (dumb idea, not an issue) Generating strings without actually adding to model.
Issue -
State: open - Opened by PenguinMan-1219 11 months ago
#75 - May be transfer_{}.py
Issue -
State: open - Opened by Alan-Qin 11 months ago
#74 - AttributeError: 'NoneType' object has no attribute 'value'
Issue -
State: open - Opened by zhu111-eng 11 months ago
#73 - Differences between demo and code
Issue -
State: closed - Opened by andyz245 11 months ago
- 1 comment
#72 - Hugging face compat
Pull Request -
State: open - Opened by dribnet 11 months ago
#71 - 24g nvidia OOM someone help!!
Issue -
State: open - Opened by daz422 11 months ago
- 10 comments
#70 - Attack success rate very low
Issue -
State: open - Opened by YerongLi 11 months ago
- 2 comments
#69 - The quality of suffixes given in the examples
Issue -
State: open - Opened by Imsovegetable 12 months ago
#68 - error: self._control_slice.start = None when using other LLMs
Issue -
State: open - Opened by aaFrostnova 12 months ago
- 4 comments
#67 - Reproducing individual behaviors
Issue -
State: open - Opened by colinpannikkat 12 months ago
#66 - About eval
Issue -
State: open - Opened by FYYFU 12 months ago
- 1 comment
#65 - SafetensorError: Error while deserializing header: HeaderTooLarge
Issue -
State: open - Opened by PengfeiHePower 12 months ago
- 1 comment
#64 - How to calculate the ASR score for harmful string and harmful behaviours
Issue -
State: open - Opened by zhao1402072392 12 months ago
#63 - No this element.
Issue -
State: open - Opened by GregxmHu 12 months ago
#62 - For other LLMs
Issue -
State: open - Opened by GregxmHu 12 months ago
#61 - Default arguments to replicate paper results
Issue -
State: open - Opened by HBSaddiq 12 months ago
- 1 comment
#60 - How can we use the result to attack other prompts?
Issue -
State: open - Opened by lizaitang about 1 year ago
#59 - Is it correct that the input_ids in the demo have two bos tokens?
Issue -
State: open - Opened by Jellyfish042 about 1 year ago
- 1 comment
#58 - Evaluation standard
Issue -
State: closed - Opened by DYR1 about 1 year ago
- 2 comments
#57 - add 8-bit demo
Pull Request -
State: open - Opened by Jellyfish042 about 1 year ago
#56 - Conflict between llama2 conv_template and AttackPrompt._update_ids() in strings case
Issue -
State: closed - Opened by ShacklesLay about 1 year ago
- 1 comment
#55 - Unable to reproduce provided Jupyter notebook results
Issue -
State: closed - Opened by suraj-srinivas about 1 year ago
- 3 comments
#54 - Potential mistake in gradients calculation ?
Issue -
State: closed - Opened by boussad83 about 1 year ago
- 1 comment
#53 - What is ARCA algorithm
Issue -
State: closed - Opened by chengjl19 about 1 year ago
- 1 comment
#52 - torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 86.00 MiB (GPU 0; 11.73 GiB total capacity; 11.04 GiB already allocated; 25.06 MiB free; 11.04 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
Issue -
State: closed - Opened by kompowiec about 1 year ago
- 3 comments
#51 - Which exact target training prompts were used for which experiments?
Issue -
State: open - Opened by j-cb about 1 year ago
#50 - Reproduce the transfer attack in the paper
Issue -
State: open - Opened by sherdencooper about 1 year ago
- 7 comments
#49 - Other baseline attacks
Issue -
State: closed - Opened by nurlanov-zh about 1 year ago
- 1 comment
#48 - The difference of the position of _target_slice and _loss_slice
Issue -
State: closed - Opened by z379035389 about 1 year ago
- 4 comments
#47 - The program is blocked while running
Issue -
State: closed - Opened by jr011 about 1 year ago
#46 - Potential mistake
Issue -
State: open - Opened by Kalyan0821 about 1 year ago
- 1 comment
#45 - Import error on colab demo
Issue -
State: closed - Opened by zczlsde about 1 year ago
- 1 comment
#44 - [reproduce] Optimization unstable / not working
Issue -
State: open - Opened by ssnl about 1 year ago
- 19 comments
#43 - Thought experiment on prevention methods
Issue -
State: closed - Opened by yhyu13 about 1 year ago
- 1 comment
#43 - Thought experiment on prevention methods
Issue -
State: closed - Opened by yhyu13 about 1 year ago
- 1 comment
#42 - Package Requirements for "demo.ipynb"
Issue -
State: closed - Opened by asif-hanif about 1 year ago
- 3 comments
#41 - GPU usage about vicuna experiments
Issue -
State: closed - Opened by hoy021 about 1 year ago
#40 - cand[-1] in get_filtered_cands of attack_manager.py may cause error
Issue -
State: open - Opened by Ethan-Chen-plus about 1 year ago
- 5 comments
#39 - The suffix is not transferable
Issue -
State: open - Opened by ziho7 about 1 year ago
- 3 comments
#38 - huggingface_hub.utils._validators.HFValidationError:
Issue -
State: open - Opened by kronkinatorix about 1 year ago
- 2 comments
#37 - evaluation error
Issue -
State: closed - Opened by adwardlee about 1 year ago
- 3 comments
#36 - GPU Resources
Issue -
State: closed - Opened by ShiriCohen about 1 year ago
- 1 comment
#35 - Possible to run on multiple GPUs?
Issue -
State: open - Opened by john-c-frank about 1 year ago
- 5 comments
#34 - Only first character of suffix is modified
Issue -
State: open - Opened by bsterrett about 1 year ago
- 4 comments
#33 - Nonetype error when reproducing experiments
Issue -
State: closed - Opened by ishcha about 1 year ago
- 3 comments
#32 - Missing folders?
Issue -
State: closed - Opened by tjaffee99 about 1 year ago
- 1 comment
#31 - RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from [link]
Issue -
State: closed - Opened by kompowiec about 1 year ago
- 1 comment
#30 - Apply GCG to Chinese-LLM
Issue -
State: open - Opened by enochdongyi about 1 year ago
- 13 comments
#29 - What's the prev_control(line:1382,attacck_manager.py)mean? Why using ''haha?
Issue -
State: closed - Opened by Ethan-Chen-plus about 1 year ago
- 2 comments
#28 - Windows filesystem does not allow ":" used in timestamp
Issue -
State: closed - Opened by SD-fan about 1 year ago
- 4 comments
#27 - Question about logging
Issue -
State: open - Opened by iamgroot42 about 1 year ago
- 7 comments
#26 - Model install step is confusing
Issue -
State: closed - Opened by tjaffee99 about 1 year ago
- 2 comments
#25 - Missing step() for MultiPromptAttack?
Issue -
State: closed - Opened by iamgroot42 about 1 year ago
- 1 comment
#24 - Possible to use quantized models? 4bit, 8bit, etc
Issue -
State: open - Opened by Ph0rk0z about 1 year ago
- 19 comments
#23 - about chinese llama tokenizer add '' at begin of chinese words
Issue -
State: closed - Opened by ckx506772099 about 1 year ago
- 3 comments
#22 - How is control weight used?
Issue -
State: closed - Opened by MaxwellJryao about 1 year ago
- 3 comments
#22 - How is control weight used?
Issue -
State: closed - Opened by MaxwellJryao about 1 year ago
- 3 comments
#21 - 'Conversation' object has no attribute 'system'
Issue -
State: closed - Opened by iamgroot42 about 1 year ago
- 7 comments
#20 - How to guard against llm-attacks?
Issue -
State: closed - Opened by AIApprentice101 about 1 year ago
- 1 comment
#19 - how to run these code in jupyter notebook?
Issue -
State: closed - Opened by Ethan-Chen-plus about 1 year ago
- 4 comments
#17 - How do control_init works?
Issue -
State: closed - Opened by Ethan-Chen-plus about 1 year ago
- 1 comment
#15 - Typo in your paper
Issue -
State: closed - Opened by claytondukes about 1 year ago
- 1 comment
#14 - Parallelism/Memory
Issue -
State: open - Opened by zack-anthropic about 1 year ago
- 14 comments
#13 - there's a typo in the arxiv paper.
Issue -
State: closed - Opened by jianghoy about 1 year ago
- 1 comment
#12 - Hugging face compat
Pull Request -
State: open - Opened by croche2574 about 1 year ago
- 1 comment
#11 - Update README.md
Pull Request -
State: closed - Opened by eltociear about 1 year ago
- 2 comments
#10 - How to slove the following problem?
Issue -
State: closed - Opened by HuichiZhou about 1 year ago
- 5 comments
#9 - A typo in paper published on arxiv
Issue -
State: closed - Opened by Melodramass about 1 year ago
- 2 comments