-
i try to stf llama2-7b and oom, can it support fsdp or tensor parallel
-
-
The prompts for Llama 2 have not been provided in `prompts/adv_prompts`, so running `load_adv_prompt` doesn't work when using Llama 2. Could these be added, please? Thanks!
-
### Model introduction
This model is created by Sharan as an experiment for my thesis. I am working on testing different Quantization and PEFT combinations.
### Model URL
https://huggingface.co/Sh…
-
This issue tracks the open issues the model team must solve in order to hit Llama2 perf targets.
## Decode 128
We have a new perf target of 20 tok/s at seqlen = 128. This issue lists the problems …
-
感谢作者的工作,提供了一个解决 cl 灾难性遗忘的思路。
我采用 codebase 提供的 llama2 的脚本,跑出来的结果直接坏掉了,这是什么原因呢,跑实验的过程中,有什么要点需要注意么,或者参数设置上需要做些什么调整呢?是 olora 的 lamda 参数设置太小导致过多的遗忘么?下面是我在 tune order2 时的逐 task 结果
***** predict metrics **…
-
Command executed .
```
guile ../../utils/sexp-json.scm < matrix_multiplication.sexp | python3 ../../c-lisp.py
```
Output received .
```
Error in statement: ['call', 'fprint', ['load', ['ptrad…
-
Hello!
Commit `2badd76` appears to break `examples.models.llama2.export_llama`, specifically with Llama 3.
### Expected Behavior
```
[INFO 2024-06-14 16:04:23,366 export_llama_lib.py:390] Ap…
amqdn updated
3 hours ago
-
## Overview
### PART 1 Downloading datasets
1. Sign up to MLcommons to access SCROLLS govreport dataset & model
1.5 Until approval is given to access dataset & model Download llama-2 fused qkv…
-
In your paper, the ASR on llama2-7b-chat using the AutoDAN method is tens of times different from the ASR given in the AutoDAN paper (although AutoDAN uses AdvBench and you use HarmBench, such a large…