Daniel Huang

VideoMAE Model Enabling and Examples

2

# What does this PR do? This PR gives examples and proves compatibility for [VideoMAE](https://huggingface.co/MCG-NJU/videomae-base-finetuned-kinetics) with Gaudi 2 on graph mode and casted to BF16. Tests included ensure compatibility with...

[Feature Request] Add new config file types

1

Currently, the config file only supports CLI-style flags as inputs. Is it possible with the current design to have the config also accept other well-known config formats such as json,...

Feature extraction enabling

1

# What does this PR do? This PR enables BERT-based models for text feature extraction. A change to `BertModel.forward` was made to avoid a NAN bug related to autocasting to...

Table transformer enabling

1

# What does this PR do? This PR enables the [table transformer detection](https://huggingface.co/microsoft/table-transformer-detection) model with one minor modification in the conv encoder module related to an indexing bug for certain...

# What does this PR do? This PR enables text to video generation for [ali-vilab/text-to-video-ms-1.7b](https://huggingface.co/ali-vilab/text-to-video-ms-1.7b) through a `GaudiTextToVideoSDPipeline` object, which was created with reference to the Stable Diffusion models. ##...

Proc-macro panicked: proc-macro server exited with signal: 6

2

**rust-analyzer version**: (eg. output of "rust-analyzer: Show RA Version" command, accessible in VSCode via Ctrl/⌘+Shift+P) 2024-08-27, installed through Mason **rustc version**: (eg. output of `rustc -V`) 1.75.0 **editor or extension**:...

Minicpm enabling

1

# What does this PR do? Enables [MiniCPM3](https://huggingface.co/openbmb/MiniCPM3-4B) model for Causal LM. Follows #1133 in optimizing remote code. The following changes were added: - `htcore.mark_step()` between each decoder layer to...

Enabling Snowflake Arctic on Gaudi 3

14

# What does this PR do? This PR enables [snowflake-arctic-instruct](https://huggingface.co/Snowflake/snowflake-arctic-instruct) on a single node in Gaudi 3. A single Gaudi 2 node with 8 cards does not have enough memory...

# What does this PR do? This PR will enable the Janus model in optimum habana. This will act as an intermediate activation of Janus model until https://github.com/huggingface/transformers/pull/36053 is supported...