-
In the file `llava/model/llava_arch.py` under the class `LlavaMetaForCausalLM` there is a function`prepare_inputs_labels_for_multimodal` that is called when calling the `generate` and `forward` functi…
-
**Is your feature request related to a problem? Please describe.**
Currently, we have the Erlang SDK running at scale in our services, tho, recently, our team start to face, what we believe to be an …
-
fatal: not a git repository (or any parent up to mount point /gpfs)
Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set).
world_size 0
root_dir ./
id_prop_csv_file exists True…
LTJer updated
3 weeks ago
-
### Description
In Llama3-70B we split 32 users across 4 device columns in galaxy for parallelizing the `FlashDecode` across batch. Then before `Selfout` we use `ttnn.line_all_gather` to concatenate …
-
Hello, I 've been trying to use the `SFTTrainer` with the [vicgalle/alpaca-gpt4](https://huggingface.co/datasets/vicgalle/alpaca-gpt4) dataset. However after prepping the dataset in the SFT format, I …
-
**Is your feature request related to a problem? Please describe.**
When doing reset, signals and updates after the reset point will be reapplied (cherry-picked) to the new run. However all those reap…
-
**Is your feature request related to a problem? Please describe.**
Current sovereign sequencer design doesn't contain any escape hatch. As a reason, if hotshot is down, batch poster will not post any…
-
Hello, I am having issues with the docker T1w segmentation not generating the corresponding pdf and jpg reports. Does anyone else have this issue? In the logs the only reference to them I am seeing is…
-
### 🐛 Describe the bug
`_add_batch_dim` and `_remove_batch_dim` are supposed to be internal functions that are not user visible. However, `__torch_function__` makes it possible to observe these funct…
-
In many batch files is a for-loop used like this
`FOR /F "EOL= tokens=* delims="`
It's used to fetch a full line and avoid skipping of lines, when they begin with the EOL character.
But usin…