The ilab model train finishing with the error: "Error: Data path must be to a valid .jsonl file. Value given: /path/to/.local/share/instructlab/datasets". #2257
Description
Describe the bug
The ilab model train finishes with the error "Expected list[str] but got tuple with value ('q_proj', 'k_proj', 'v_proj', 'o_proj') - serialized value may not be as expected".
To Reproduce
Steps to reproduce the behavior:
- Run the
ilab model train. - See the following error. The error message is similar with the one on the Click converts lists into tuples when parsing args #2061. But the error message is a bit different.
(venv) $ ilab model train
/home/jaruga/doc/dev/instructlab/venv/lib64/python3.11/site-packages/pydantic/main.py:387: UserWarning: Pydantic serializer warnings:
Expected `list[str]` but got `tuple` with value `('q_proj', 'k_proj', 'v_proj', 'o_proj')` - serialized value may not be as expected
return self.__pydantic_serializer__.to_python(
Usage: ilab model train [OPTIONS]
Try 'ilab model train --help' for help.
Error: Data path must be to a valid .jsonl file. Value given: /home/jaruga/.local/share/instructlab/datasets
Expected behavior
The ilab model train finishes without an error.
Screenshots
Device Info (please complete the following information):
- Hardware Specs: [e.g. Apple M2 Pro Chip, 16 GB Memory, etc.]
- OS Version: [e.g. Mac OS 14.4.1, Fedora Linux 40]
- Python Version: [output of
python --version] - InstructLab Version: [output of
ilab system info]
Hardware Specs:
$ neofetch --stdout
jaruga@fedora
-------------
OS: Fedora release 39 (Thirty Nine) x86_64
Host: 20KGS23S0P ThinkPad X1 Carbon 6th
Kernel: 6.9.12-100.fc39.x86_64
Uptime: 14 days, 3 hours, 10 mins
Packages: 3791 (rpm), 21 (brew)
Shell: bash 5.2.26
Resolution: 1920x1080
WM: sway
Theme: Adwaita [GTK2/3]
Icons: Adwaita [GTK2/3]
Terminal: alacritty
CPU: Intel i7-8650U (8) @ 4.200GHz
GPU: Intel UHD Graphics 620
Memory: 5854MiB / 15635MiB
OS Version: Fedora Linux 39
Python Version:
(venv) $ python --version
Python 3.11.9
$ rpm -q python3.11
python3.11-3.11.9-6.fc39.x86_64
InstructLab Version:
(venv) $ ilab system info
sys.version: 3.11.9 (main, Aug 23 2024, 00:00:00) [GCC 13.3.1 20240522 (Red Hat 13.3.1-1)]
sys.platform: linux
os.name: posix
platform.release: 6.9.12-100.fc39.x86_64
platform.machine: x86_64
platform.node: localhost.localdomain
platform.python_version: 3.11.9
os-release.ID: fedora
os-release.VERSION_ID: 39
os-release.PRETTY_NAME: Fedora Linux 39 (Workstation Edition)
instructlab.version: 0.18.4
instructlab-dolomite.version: 0.1.1
instructlab-eval.version: 0.1.2
instructlab-quantize.version: 0.1.0
instructlab-schema.version: 0.3.1
instructlab-sdg.version: 0.2.7
instructlab-training.version: 0.4.2
torch.version: 2.3.1+cu121
torch.backends.cpu.capability: AVX2
torch.version.cuda: 12.1
torch.version.hip: None
torch.cuda.available: False
torch.backends.cuda.is_built: True
torch.backends.mps.is_built: False
torch.backends.mps.is_available: False
llama_cpp_python.version: 0.2.79
llama_cpp_python.supports_gpu_offload: False
Additional context
I ran the following commands before running the ilab model train.
(venv) $ ilab data generate --pipeline simple
...
INFO 2024-09-12 14:03:35,328 instructlab.sdg.datamixing:200: Mixed Dataset saved to /home/jaruga/.local/share/instructlab/datasets/skills_train_msgs_2024-09-12T13_05_40.jsonl
INFO 2024-09-12 14:03:35,329 instructlab.sdg:438: Generation took 3474.95s
(venv) $ ilab taxonomy diff
compositional_skills/miscellaneous_unknown/qna.yaml
Taxonomy in /home/jaruga/.local/share/instructlab/taxonomy is valid :)
(venv) $ ilab model list
+------------------------------+---------------------+--------+
| Model Name | Last Modified | Size |
+------------------------------+---------------------+--------+
| merlinite-7b-lab-Q4_K_M.gguf | 2024-09-09 18:44:12 | 4.1 GB |
+------------------------------+---------------------+--------+
The used pip package versions.
(venv) $ pip list | grep instructlab
instructlab 0.18.4
instructlab-dolomite 0.1.1
instructlab-eval 0.1.2
instructlab-quantize 0.1.0
instructlab-schema 0.3.1
instructlab-sdg 0.2.7
instructlab-training 0.4.2
(venv) $ pip list | grep pydantic
pydantic 2.9.1
pydantic_core 2.23.3
pydantic-settings 2.4.0
pydantic_yaml 1.3.0
The pydantic version 2.9.1 is used. So, the error line pydantic/main.py:387 is here.
(venv) $ ilab model train /home/jaruga/doc/dev/instructlab/venv/lib64/python3.11/site-packages/pydantic/main.py:387: > UserWarning: Pydantic serializer warnings: Expected `list[str]` but got `tuple` with value `('q_proj', 'k_proj', 'v_proj', 'o_proj')` - serialized value may not be as expected return self.__pydantic_serializer__.to_python( Usage: ilab model train [OPTIONS] Try 'ilab model train --help' for help.
In addition, I am not sure if this is relevant to this issue. I ran the git pull command manually after setting up the /home/jaruga/.local/share/instructlab/taxonomy by the ilab config init.
$ pwd
/home/jaruga/.local/share/instructlab/taxonomy
$ git pull