#First forum error WSL conda install from guide

57 messages · Page 1 of 1 (latest)

vast mantle
#

(unsloth) ed@EdWindows10:~$ python -m unsloth
Traceback (most recent call last):
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/runpy.py", line 187, in _run_module_as_main
mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/runpy.py", line 146, in _get_module_details
return _get_module_details(pkg_main_name, error)
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/runpy.py", line 110, in _get_module_details
import(pkg_name)
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/unsloth/init.py", line 112, in <module>
from .models import *
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/unsloth/models/init.py", line 15, in <module>
from .loader import FastLanguageModel
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/unsloth/models/loader.py", line 15, in <module>
from .llama import FastLlamaModel, logger
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/unsloth/models/llama.py", line 18, in <module>
from transformers.models.llama.modeling_llama import (
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/transformers/models/llama/modeling_llama.py", line 53, in <module>
from flash_attn import flash_attn_func, flash_attn_varlen_func
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/flash_attn/init.py", line 3, in <module>
from flash_attn.flash_attn_interface import (
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 10, in <module>
import flash_attn_2_cuda as flash_attn_cuda
ImportError: libcudart.so.12: cannot open shared object file: No such file or directory

#

trying the nightly from the github : pip install "unsloth[kaggle-new] @ git+https://github.com/unslothai/unsloth.git@nightly"

#

same error

#

think updating cuda would help?

#

(unsloth) ed@EdWindows10:~$ python -m triton
/home/ed/mambaforge/envs/unsloth/bin/python: No module named triton.main; 'triton' is a package and cannot be directly executed

#

thinking it may be cos no triton

#

nope that wasnt it

hard wren
vast mantle
#

(unsloth) ed@EdWindows10:~$ ln -s /usr/lib/wsl/lib/libcuda.so /home/ed/anaconda3/envs/sglang_3_11_7/lib/libcuda.so
ln: failed to create symbolic link '/home/ed/anaconda3/envs/sglang_3_11_7/lib/libcuda.so': No such file or directory

#

oh wait lol

#

ignore me

#

(unsloth) ed@EdWindows10:~$ ln -s /usr/lib/wsl/lib/libcuda.so /home/ed/anaconda3/envs/unsloth/lib/libcuda.so
ln: failed to create symbolic link '/home/ed/anaconda3/envs/unsloth/lib/libcuda.so': No such file or directory

#

oh wait no still nothing sorry

#

aha again sorry im using mambaforge

#

will have another try

#

yeah still nothing

#

(unsloth) ed@EdWindows10:~$ python -m unsloth
Traceback (most recent call last):
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/runpy.py", line 187, in _run_module_as_main
mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/runpy.py", line 146, in _get_module_details
return _get_module_details(pkg_main_name, error)
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/runpy.py", line 110, in _get_module_details
import(pkg_name)
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/unsloth/init.py", line 112, in <module>
from .models import *
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/unsloth/models/init.py", line 15, in <module>
from .loader import FastLanguageModel
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/unsloth/models/loader.py", line 15, in <module>
from .llama import FastLlamaModel, logger
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/unsloth/models/llama.py", line 18, in <module>
from transformers.models.llama.modeling_llama import (
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/transformers/models/llama/modeling_llama.py", line 53, in <module>
from flash_attn import flash_attn_func, flash_attn_varlen_func
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/flash_attn/init.py", line 3, in <module>
from flash_attn.flash_attn_interface import (
File "/home/ed/mambaforge/envs/unsloth/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 10, in <module>
import flash_attn_2_cuda as flash_attn_cuda
ImportError: libcudart.so.12: cannot open shared object file: No such file or directory

hard wren
vast mantle
#

where should that be stored>

#

cant see it in the wsl/lib/ folder

#

you using 12 or 11 for cuda?

hard wren
vast mantle
#

ill upgrade

patent shale
#

even pip install torch torchvision torchaudio from the official Pytorch docs does not work?

#

after torch, I normally go into python and type import torch; torch.__version__; and also torch.version.cuda or was it torch.cuda.version

#

if torch succeeds, i then go do pip install -U xformers --index-url https://download.pytorch.org/whl/cu121 then try python -m xformers.info to see if theres issues

#

ie check if all avaliable

#

if that succeeds then pip install bitsandbytes

#

then do python -m bitsandbytes

#

then finally do pip install "unsloth[kaggle-new] @ git+https://github.com/unslothai/unsloth.git@nightly"

vast mantle
#

retrying with 12.1 now thanks for all the help so far guys

patent shale
#

oh great 12.1 works! Thanks to @hard wren for mentioning it!!

hard wren
vast mantle
#

oh not got it running yet still installing stuff

#

aha sorry

#

import torch; torch.version;
'2.1.2'

patent shale
#

OHH loll my wording and punctuation is wrong lmao

#

it should be "Oh! Great 12.1 works for you!"

#

lolll whoops

vast mantle
#

@hard wren if i cant get it working this time think you could share your yaml for the env?

hard wren
#

I never needed to use it, all I ever did was make symlinks

vast mantle
#

memory_efficient_attention.tritonflashattF: unavailable
memory_efficient_attention.tritonflashattB: unavailable

#

sequence_parallel_fused.write_values: unavailable
sequence_parallel_fused.wait_values: unavailable
sequence_parallel_fused.cuda_memset_32b_async: unavailable

#

dcgm_profiler: unavailable

#

continuing with the rest anyway lets see what happends

#

config.json: 100%|█████████████████████████████████████████████████████████████████| 1.11k/1.11k [00:00<00:00, 9.36MB/s]
==((====))== Unsloth: Fast Gemma patching release 2024.3
\ /| GPU: NVIDIA GeForce RTX 3090. Max memory: 23.999 GB. Platform = Linux.
O^O/ _/ \ Pytorch: 2.2.0+cu121. CUDA = 8.6. CUDA Toolkit = 12.1.
\ / Bfloat16 = TRUE. Xformers = 0.0.24. FA = False.
"-____-" Free Apache license: http://github.com/unslothai/unsloth
/home/ed/mambaforge/envs/unsloth_env/lib/python3.10/site-packages/transformers/quantizers/auto.py:155: UserWarning: You passed quantization_config or equivalent parameters to from_pretrained but the model you're loading already has a quantization_config attribute. The quantization_config from the model will be used.
warnings.warn(warning_msg)
model.safetensors: 4%|██▍ | 231M/5.57G [00:09<04:02, 22.0MB/s]

GitHub
GitHub - unslothai/unsloth: 5X faster 60% less memory QLoRA finetuning

5X faster 60% less memory QLoRA finetuning. Contribute to unslothai/unsloth development by creating an account on GitHub.

#

wooo

#

seeming like 11.8 cuda was the issue

#

ModuleNotFoundError: No module named 'click' can probs fix but thought i would pop here so otheres know to expect it

#

after pip install click i got ModuleNotFoundError: No module named 'google.protobuf'

#

trying conda install protobuf

#

mamba worked better

#

oh wait no it did not

#

restarting seems to have fixes it training run started

#

sorted model uploading now cheers guys