Fanli Lin

predict function doesn't return the correct predictions with num_actors >1

6

Hi, when using the following code snippet to do xgboost training, I noticed that the results that the `predict` function returns are different when I set the number of actors...

feat: enable loading tokenizer from YAML file for models not supported in transformers

5

### Related Issues - fixes [#issue-number](https://github.com/deepset-ai/haystack/issues/5313) ### Proposed Changes: In the `_prepare_pipeline_kwargs` method, we can check whether `transformers` would load the tokenizer if we pass it as a string. If...

fix: enable passing more parameters to the generation model

2

### Related Issues N/A ### Proposed Changes: When running the following code using PromptNode, I noticed that the defined `generate_kwargs` passed to `prompt_node.run()` are not successfully passed down to the...

**Problem** I want to load the following pipeline defined in YAML files into Haystack and run a query to get some results. This is my YAML file: ```python version: 1.19.0rc0...

style: `max_new_tokens` are set twice with the same value

1

### Related Issues In the following code snippet, we can see that for text-generation task, `max_new_tokens` are set twice: first set to self.max_length and then overwritten by `max_length` or self.max_length....

enable Pipeline to get device from model

5

## What does this PR do? ```python import torch from transformers import AutoModelForCausalLM from transformers import AutoTokenizer, pipeline model_id = "meta-llama/Llama-2-7b-chat-hf" model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda") print(model.device) tokenizer = AutoTokenizer.from_pretrained(model_id) pipe...

can't pass an object parameter to pipeline.run()

10

**Describe the bug** I implemented a FastAPITokenStreamingHandler object in the RESTAPI layer following the example mentioned in [this](https://github.com/deepset-ai/haystack/pull/4889) PR and pass it as a parameter to `pipeline.run()`. The only difference...

[tests] fix deepspeed zero3 config for `test_stage3_nvme_offload`

1

## What does this PR do? Since we manually modified the original zero3 config value [here](https://github.com/huggingface/transformers/blob/main/tests/deepspeed/test_deepspeed.py#L402), we will end up with a ValueError in accelerate ([code](https://github.com/huggingface/accelerate/blob/main/src/accelerate/accelerator.py#L3269)). For Zero3 Checkpointing, we...

add IPEX-XPU support for Llama2 model Inference

3

## What does this PR do? This PR enables Intel GPU support for Llama2 model inference in optimum-intel. Below is a code example: ```python import torch from transformers import AutoTokenizer,...

## What does this PR do? As can be seen from the fix, the correct variable names for tensorflow and pytorch should be used. cc @stevhliu