-
I'm trying to fit the Hybrid Fastconformer for streaming, and found a strange thing, in the config that you specify with huggingface, it is indicated that self_attention_model: rel_pos. However, in th…
-
Hi, I am using llm as part of a multimodal model, so the model needs to pass `input embedding tensor` directly to generate, and also need to access the language model's `embed_tokens` member to fist c…
-
Loading checkpoint shards: 0%| | 0/5 [00:00
-
# 🐛 Bug
from vllm import LLM, SamplingParams
llm = LLM(model=model_dir,enforce_eager=True)
then
```
File d:\my\env\python3.10.10\lib\site-packages\xformers\ops\fmha\_triton\splitk_kernels.…
-
**When I run these two, I get this error: RuntimeError: The shape of the 2D attn_mask is torch.Size([77, 77]), but should be (1, 1). Specific errors are as follows:**
F:\Miniconda\envs\dream\lib\site…
-
RuntimeError Traceback (most recent call last)
Cell In[5], line 10
5 prompt = "On Christmas evening, on a crowded sidewalk, this item sits on the road, covered in …
-
I want to run BundleSDF on custom data of high resolution (e.g. 720x1080). If I follow [this instruction](https://github.com/NVlabs/BundleSDF?tab=readme-ov-file#run-on-your-custom-data), I face the fo…
-
With the decoupling of encoders and decoders, we have added a `Linear` encoder, which seems to just embed the inputs and pass them along. We should also add a `SelfAttention` encoder, which encodes th…
-
I tried to load Lora training adapters from Deepspeed checkpoint:
dir:
```
ls Bunny/checkpoints-llama3-8b/bunny-lora-llama3-8b-attempt2/checkpoint-6000
total 696M
-rw-r--r-- 1 schwan46494@gmail.c…
-
感谢作者在可解释性方面做出的优秀工作。我目前在用llama-3.1-8b做一些研究，在给transformer_lens中添加了`meta-llama/Llama-3.1-8B-Instruct`的支持代码后，运行`knowledge_eap.ipynb`，发现在第6个单元格计算`attribute(model, g, data, partial(logit_diff, loss=True, me…