Jannis Schönleber comments

Results 24 comments of


                                            Jannis Schönleber

Integrate openTitan `virtual` keyword (proposal)

Thank you for clarifying 👍

tree-unwrapper: paren_group not found situation detected

Same error can happen when one has something like this: ```sv `ifdef VERILATOR initial begin $display("Latch-based register file not supported for Verilator simulation"); $fatal; // works with $fatal(); end `endif...

GPTQ & AWQ Fused MOE

What is the merge plan here?

Load Mixtral 8x7b AWQ model failed

For me, both `AWQ` and `GPTQ` load, but `AWQ` just produces zero tokens as output. Commands used: ```bash # run on 2x 4090GTX CUDA_VISIBLE_DEVICES=6,7 python -m vllm.entrypoints.openai.api_server --model="TheBloke/Mixtral-8x7B-Instruct-v0.1-GPTQ" --quantization gptq...

Load Mixtral 8x7b AWQ model failed

> > For me, both `AWQ` and `GPTQ` load, but `AWQ` just produces zero tokens as output. > > Commands used: > > ```shell > > # run on 2x...

Load Mixtral 8x7b AWQ model failed

> Hi @joennlae, thanks a lot for noticing this issue with the tokenizer and the prediction of the number `2`. I am facing this issue too. Did you manage to...

systemverilog-plugin: add parameter type propagation through hierarchy

I ran the `clang-format` :-).

vLLM 0.4.0, 30% base image size reduction

Absolut banger of a pull request :-)

Train and test on model level

Hey :-) Thank you for reaching out. For example, the `Resnet9` model can be found here: https://github.com/joennlae/halutmatmul/blob/4655152246e4a7203a9401bd6f4a71905e9cb682/src/python/models/resnet9.py#L43 So, for the default pre-training, you could run the standard training script: ```bash...

Train and test on model level

I see how this can be confusing. The prototype is used when we run it like in the LUT-NN paper (https://arxiv.org/abs/2302.03213). Back in the day, there was no implementation of...