Jiao Wang

PPML document needs to be refined and corrected

15

PPML user guide is not clear and lack of some essential information. And also there are errors for the scripts. For example, For 2.1 it runs ./ppml/scripts/generate-keys.sh, it will run...

tf2 estimator failed with horovod backend if data_creator is tf.data.Dataset from generator

20

TF2Estimator failed with horovod backend if data_creator is tf.data.Dataset from generator, the error is as below: 2021-07-21 04:17:10,169 WARNING worker.py:1107 -- A worker died or was killed while executing task...

TF2Estimator prediction hangs on cluster mode

5

TF2Estimator prediction hangs on cluster mode, no display after prediction info of tensorflow. It works on local mode.

Fix to support portable zip on transformers 4.36. Related issue: https://github.com/analytics-zoo/nano/issues/1042

The repo "internlm/internlm-chat-7b-8k" does not exist now. So do the corresponding changes from "internlm/internlm-chat-7b-8k" to "internlm/internlm-chat-7b". Verify the gpu/cpu examples of repo "internlm/internlm-chat-7b".

Add a custom layer at the end of a trained Model

4

From user group issue, how to add layer to end of existing dllib keras graph model. Using nodes(someNames) or node(someName) can only apply to layers having name. The getInputs() and...

Add unittest of bigdl llm for transformer 4.36 - llama - bloom - chatglm - replit_code - mistral - whisper - falcon - mpt

Fix issue: https://github.com/intel-analytics/ipex-llm/issues/10661 Also correct README. Also fix all print error message issue.

chatglm cannot work on transformers 4.34.1+ due to incompatible tokenizer, see https://github.com/analytics-zoo/nano/issues/917#issuecomment-1888268638 Update readme to install transformers < 4.34.1

Fix issue of running LLaVA example on transformers 4.36. Related issue: https://github.com/analytics-zoo/nano/issues/1143 The fix includes: In llama_model_forward_4_36, get real input either from input_ids or inputs_embeds , and use this input...