ImportError: cannot import name 'is_bfloat16_supported' from 'unsloth' on local GPU

[regstuff]

Using the latest unsloth I get this error when trying to do a lora finetune of Mistral 7B.

Inferencing works just fine btw.

🦥 Unsloth: Will patch your computer to enable 2x faster free finetuning.
==((====))==  Unsloth: Fast Mistral patching release 2024.5
   \\   /|    GPU: NVIDIA A2. Max memory: 14.542 GB. Platform = Linux.
O^O/ \_/ \    Pytorch: 2.3.0+cu121. CUDA = 8.6. CUDA Toolkit = 12.1.
\        /    Bfloat16 = TRUE. Xformers = 0.0.26.post1. FA = True.
 "-____-"     Free Apache license: http://github.com/unslothai/unsloth
Unsloth 2024.5 patched 32 layers with 32 QKV layers, 32 O layers and 32 MLP layers.
Traceback (most recent call last):
  File "/home/sravanth/unsloth/finetune_unsloth.py", line 66, in <module>
    from unsloth import is_bfloat16_supported
ImportError: cannot import name 'is_bfloat16_supported' from 'unsloth' (/home/sravanth/unsloth/unsloth/lib/python3.10/site-packages/unsloth/__init__.py)

Here's the relevant code block. This code works fine in colab though. (JFYI my local install is done as per the github repo instructions. I did not simply copy the colab install cells.)

from unsloth import FastLanguageModel
import torch
max_seq_length = 8192 # Choose any! We auto support RoPE Scaling internally!
dtype = None # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+
load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False.

# 4bit pre quantized models we support for 4x faster downloading + no OOMs.
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "unsloth/mistral-7b-bnb-4bit", # Choose ANY! eg teknium/OpenHermes-2.5-Mistral-7B
    max_seq_length = max_seq_length,
    dtype = dtype,
    load_in_4bit = load_in_4bit,
    # token = "hf_...", # use one if using gated models like meta-llama/Llama-2-7b-hf
)

model = FastLanguageModel.get_peft_model(
    model,
    r = 32, # Choose any number > 0 ! Suggested 8, 16, 32, 64, 128
    target_modules = ["q_proj", "k_proj", "v_proj", "o_proj",
                      "gate_proj", "up_proj", "down_proj",],
    lora_alpha = 64,
    lora_dropout = 0, # Supports any, but = 0 is optimized
    bias = "none",    # Supports any, but = "none" is optimized
    # [NEW] "unsloth" uses 30% less VRAM, fits 2x larger batch sizes!
    use_gradient_checkpointing = "unsloth", # True or "unsloth" for very long context
    random_state = 3407,
    use_rslora = False,  # We support rank stabilized LoRA
    loftq_config = None, # And LoftQ
)

from trl import SFTTrainer
from transformers import TrainingArguments
from unsloth import is_bfloat16_supported

trainer = SFTTrainer(
    model = model,
    tokenizer = tokenizer,
    train_dataset = train_dataset,
    eval_dataset = eval_dataset,
    dataset_text_field = "text",
    max_seq_length = max_seq_length,
    dataset_num_proc = 2,
    packing = False, # Can make training 5x faster for short sequences.
    args = TrainingArguments(
        per_device_train_batch_size = 1,
        gradient_accumulation_steps = 8,
        warmup_steps = 5,
        num_train_epochs=2,
        learning_rate = 2e-4,
        fp16 = not is_bfloat16_supported(),
        bf16 = is_bfloat16_supported(),
        logging_steps = 1,
        optim = "adamw_8bit",
        weight_decay = 0.01,
        lr_scheduler_type = "linear",
        seed = 3407,
        output_dir = "outputs",
        fp16_full_eval = True,
        per_device_eval_batch_size = 1,
        eval_accumulation_steps = 8,
        evaluation_strategy = "steps",
        eval_steps = 1,
    ),
)

trainer_stats = trainer.train()

[danielhanchen]

@regstuff Could you try uninstalling then installing Unsloth ie:

pip uninstall unsloth -y
pip install --upgrade --force-reinstall --no-cache-dir git+https://github.com/unslothai/unsloth.git

[regstuff]

@danielhanchen
Thanks. Deleted my whole virtual env and did everything from scratch, and it's working now. For some weir reason I also have to install packaging via pip otherwise it says can't find this module.

[danielhanchen]

Oh interesting on packaging