+
+
+
+| + +## Table of Contents +- [Introduction](#axolotl) +- [Supported Features](#axolotl-supports) +- [Quickstart](#quickstart-) +- [Environment](#environment) + - [Docker](#docker) + - [Conda/Pip venv](#condapip-venv) + - [Cloud GPU](#cloud-gpu) - Latitude.sh, JarvisLabs, RunPod + - [Bare Metal Cloud GPU](#bare-metal-cloud-gpu) + - [Windows](#windows) + - [Mac](#mac) + - [Google Colab](#google-colab) + - [Launching on public clouds via SkyPilot](#launching-on-public-clouds-via-skypilot) + - [Launching on public clouds via dstack](#launching-on-public-clouds-via-dstack) +- [Dataset](#dataset) +- [Config](#config) + - [Train](#train) + - [Inference](#inference-playground) + - [Merge LORA to Base](#merge-lora-to-base) + - [Special Tokens](#special-tokens) + - [All Config Options](#all-config-options) +- Advanced Topics + - [Multipack](./docs/multipack.qmd) + - [RLHF & DPO](./docs/rlhf.qmd) + - [Dataset Pre-Processing](./docs/dataset_preprocessing.qmd) +- [Common Errors](#common-errors-) + - [Tokenization Mismatch b/w Training & Inference](#tokenization-mismatch-bw-inference--training) +- [Debugging Axolotl](#debugging-axolotl) +- [Need Help?](#need-help-) +- [Badge](#badge-) +- [Community Showcase](#community-showcase) +- [Contributing](#contributing-) +- [Sponsors](#sponsors-) + + | +
+
+
+
+
+
+
+
+
+ Axolotl provides a unified repository for fine-tuning + Go ahead and Axolotl questions!! + +
+ 
+ |
+
](https://github.com/OpenAccess-AI-Collective/axolotl)
+```
+
+[](https://github.com/OpenAccess-AI-Collective/axolotl)
+
+## Community Showcase
+
+Check out some of the projects and models that have been built using Axolotl! Have a model you'd like to add to our Community Showcase? Open a PR with your model.
+
+Open Access AI Collective
+- [Minotaur 13b](https://huggingface.co/openaccess-ai-collective/minotaur-13b-fixed)
+- [Manticore 13b](https://huggingface.co/openaccess-ai-collective/manticore-13b)
+- [Hippogriff 30b](https://huggingface.co/openaccess-ai-collective/hippogriff-30b-chat)
+
+PocketDoc Labs
+- [Dan's PersonalityEngine 13b LoRA](https://huggingface.co/PocketDoc/Dans-PersonalityEngine-13b-LoRA)
+
+## Contributing 🤝
+
+Please read the [contributing guide](./.github/CONTRIBUTING.md)
+
+Bugs? Please check the [open issues](https://github.com/OpenAccess-AI-Collective/axolotl/issues/bug) else create a new Issue.
+
+PRs are **greatly welcome**!
+
+Please run the quickstart instructions followed by the below to setup env:
+```bash
+pip3 install -r requirements-dev.txt -r requirements-tests.txt
+pre-commit install
+
+# test
+pytest tests/
+
+# optional: run against all files
+pre-commit run --all-files
+```
+
+Thanks to all of our contributors to date. Help drive open source AI progress forward by contributing to Axolotl.
+
+
+ 
diff --git a/docs/nccl.qmd b/docs/nccl.qmd
new file mode 100644
index 0000000000000000000000000000000000000000..3b616aa66526e1d18cc3b292a7261c6470ce6540
--- /dev/null
+++ b/docs/nccl.qmd
@@ -0,0 +1,49 @@
+---
+title: NCCL
+description: Troubleshooting NCCL issues
+---
+
+NVIDIA NCCL is a library to facilitate and optimize multi-GPU communication operations, such as broadcast, all-gather, reduce, all-reduce, etc. Broadly, NCCL configuration is highly environment-specific and is configured via several [environment variables](https://docs.nvidia.com/deeplearning/nccl/user-guide/docs/env.html). A common NCCL-related problem occurs when a long-running operation times out causing the training process to abort:
+
+```text
+Watchdog caught collective operation timeout: WorkNCCL(SeqNum=42, OpType=ALLGATHER, Timeout(ms)=1800000) ran for 1806948 milliseconds before timing out.
+```
+
+Often, this timeout will happen after 30 minutes (the default setting) and is accompanied by below-average power consumption with near 100% GPU utilization before the error is raised. Nvidia recommends [disabling PCI access control services (ACS)](https://docs.nvidia.com/deeplearning/nccl/user-guide/docs/troubleshooting.html#pci-access-control-services-acs) as a possible solution if this is available to you.
+
+Forcing cross-GPU communication via [NVLink](https://en.wikipedia.org/wiki/NVLink) may help without increasing timeouts. To verify that your configuration is leveraging NVLink run the following command:
+
+```shell
+nvidia-smi nvlink --status
+```
+
+To force NCCL to use NVLink, simply set this in the environment:
+
+```shell
+export NCCL_P2P_LEVEL=NVL
+```
+
+If NVLink is not available in your environment there are other options for ``NCCL_P2P_LEVEL`` in the table below:
+
+| NCCL_P2P_LEVEL | Description |
+| -------------- | ----------- |
+| PIX | P2P data transfers through no more than a single PCIe bridge. Faster data transfer rates vs to paths involving multiple bridges, but slower compared to direct GPU-to-GPU communication. |
+| PXB | P2P data transfers through multiple PCIe bridges but not going through the PCIe Host Bridge; this path involves a complex routing process, potentially incurring a moderate level of latency. |
+| PHB | P2P data transfers occur over the PCIe and through a PCIe Host Bridge, typically involving the CPU, which can facilitate direct memory access but might introduce additional latency compared to more direct paths (ex PIX, NVL) |
+
+To validate that acceptable data transfer speeds exist for your training job, running [NCCL Tests](https://github.com/NVIDIA/nccl-tests/blob/master/README.md) can help pinpoint bottlenecks, for example:
+
+```shell
+./build/all_reduce_perf -b 8 -e 128M -f 2 -g 3
+```
+
+It can be useful when debugging NCCL communication timeouts to activate additional logging in both PyTorch and NCCL:
+
+```shell
+export NCCL_DEBUG=INFO
+export NCCL_DEBUG_SUBSYS=ALL
+export TORCH_DISTRIBUTED_DEBUG=INFO
+export TORCHELASTIC_ERROR_FILE=/PATH/TO/torcherror.log
+```
+
+Finally, if you believe your training job needs more time you can increase the timeout past 30 minutes by setting the ``ddp_timeout`` value in the Axolotl configuration. See [PyTorch init_process_group](https://pytorch.org/docs/stable/distributed.html#torch.distributed.init_process_group) for documentation on this value.
diff --git a/docs/rlhf.qmd b/docs/rlhf.qmd
new file mode 100644
index 0000000000000000000000000000000000000000..b8b2bded09a6c67a9609218db8b6aa4f394cbbe0
--- /dev/null
+++ b/docs/rlhf.qmd
@@ -0,0 +1,72 @@
+---
+title: "RLHF (Beta)"
+description: "Reinforcement Learning from Human Feedback is a method whereby a language model is optimized from data using human feedback."
+---
+
+### Overview
+
+Reinforcement Learning from Human Feedback is a method whereby a language model is optimized from data using human
+feedback. Various methods include, but not limited to:
+
+- Proximal Policy Optimization (PPO) (not yet supported in axolotl)
+- Direct Preference Optimization (DPO)
+- Identity Preference Optimization (IPO)
+
+
+### RLHF using Axolotl
+
+>[!IMPORTANT]
+>This is a BETA feature and many features are not fully implemented. You are encouraged to open new PRs to improve the integration and functionality.
+
+The various RL training methods are implemented in trl and wrapped via axolotl. Below are various examples with how you can use various preference datasets to train models that use ChatML
+
+#### DPO
+```yaml
+rl: dpo
+datasets:
+ - path: Intel/orca_dpo_pairs
+ split: train
+ type: chatml.intel
+ - path: argilla/ultrafeedback-binarized-preferences
+ split: train
+ type: chatml.argilla
+```
+
+#### IPO
+```yaml
+rl: ipo
+```
+
+#### ORPO
+
+Paper: https://arxiv.org/abs/2403.07691
+
+```yaml
+rl: orpo
+orpo_alpha: 0.1
+remove_unused_columns: false
+
+chat_template: chatml
+datasets:
+ - path: argilla/ultrafeedback-binarized-preferences-cleaned
+ type: chat_template.argilla
+```
+
+#### Using local dataset files
+```yaml
+datasets:
+ - ds_type: json
+ data_files:
+ - orca_rlhf.jsonl
+ split: train
+ type: chatml.intel
+```
+
+#### Trl autounwrap for peft
+
+Trl supports autounwrapping peft models, so that a ref model does not need to be additionally loaded, leading to less VRAM needed. This is on by default. To turn it off, pass the following config.
+
+```yaml
+# load ref model when adapter training.
+rl_adapter_ref_model: true
+```
diff --git a/examples/cerebras/btlm-ft.yml b/examples/cerebras/btlm-ft.yml
new file mode 100644
index 0000000000000000000000000000000000000000..ba4e65daaef2149c3269d85a8eb3ef3ecb7a14ff
--- /dev/null
+++ b/examples/cerebras/btlm-ft.yml
@@ -0,0 +1,89 @@
+base_model: cerebras/btlm-3b-8k-base
+model_type: AutoModelForCausalLM
+tokenizer_type: GPT2Tokenizer
+trust_remote_code: true
+tokenizer_use_fast: true
+tokenizer_legacy: true
+
+load_in_8bit: false
+load_in_4bit: false
+strict: false
+push_dataset_to_hub:
+hf_use_auth_token: true
+datasets:
+ - path: mhenrichsen/alpaca_2k_test
+ type: alpaca
+dataset_prepared_path: last_prepared_run
+val_set_size: 0.05
+
+adapter:
+lora_model_dir:
+sequence_len: 2048
+max_packed_sequence_len:
+sample_packing: false
+sample_packing_eff_est:
+sample_packing_seq_len_multiplier:
+total_num_tokens:
+
+lora_r:
+lora_alpha:
+lora_dropout:
+lora_target_modules:
+lora_target_linear:
+lora_fan_in_fan_out:
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+output_dir: ./outputs/btlm-out
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 1
+optimizer: adamw_torch
+adam_beta2: 0.95
+adam_eps: 0.000000001
+max_grad_norm: 1.0
+
+torchdistx_path:
+lr_scheduler: cosine
+lr_quadratic_warmup: true
+learning_rate: 0.000085
+train_on_inputs: true
+group_by_length: false
+bf16: auto
+fp16:
+tf32: true
+
+gradient_checkpointing: false
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+
+xformers_attention:
+flash_attention: true
+sdp_attention:
+flash_optimum:
+
+gptq_groupsize:
+gptq_model_v1:
+
+warmup_steps: 32
+evals_per_epoch: 4
+saves_per_epoch: 1
+save_total_limit:
+
+debug:
+deepspeed:
+weight_decay: 0.1
+special_tokens:
+ pad_token: "<|endoftext|>"
+fsdp:
+# - full_shard
+# - auto_wrap
+fsdp_config:
+# fsdp_state_dict_type: FULL_STATE_DICT
+# fsdp_transformer_layer_cls_to_wrap: BTLMBlock
diff --git a/examples/cerebras/qlora.yml b/examples/cerebras/qlora.yml
new file mode 100644
index 0000000000000000000000000000000000000000..285607a4c81aff6141d3cca982c6810f48e52d35
--- /dev/null
+++ b/examples/cerebras/qlora.yml
@@ -0,0 +1,59 @@
+base_model: cerebras/Cerebras-GPT-1.3B
+load_in_8bit: false
+load_in_4bit: true
+strict: false
+push_dataset_to_hub:
+datasets:
+ - path: teknium/GPT4-LLM-Cleaned
+ type: alpaca
+dataset_prepared_path:
+val_set_size: 0.05
+adapter: qlora
+lora_model_dir:
+sequence_len: 2048
+lora_r: 16
+lora_alpha: 32
+lora_dropout: 0.05
+lora_target_modules:
+ - c_fc
+ - c_attn
+ - c_proj
+lora_target_linear:
+lora_fan_in_fan_out:
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+output_dir: ./outputs/qlora-out
+batch_size: 4
+micro_batch_size: 4
+num_epochs: 2
+optimizer: paged_adamw_8bit
+torchdistx_path:
+lr_scheduler: cosine
+learning_rate: 0.0002
+train_on_inputs: false
+group_by_length: false
+bf16: auto
+fp16:
+tf32: true
+gradient_checkpointing: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention: true
+flash_attention:
+gptq_groupsize:
+gptq_model_v1:
+warmup_steps: 10
+evals_per_epoch: 4
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.1
+fsdp:
+fsdp_config:
+special_tokens:
+ pad_token: "<|endoftext|>"
diff --git a/examples/code-llama/13b/lora.yml b/examples/code-llama/13b/lora.yml
new file mode 100644
index 0000000000000000000000000000000000000000..0ba96cfaa7410a004d3afd043c87b42c27682a4a
--- /dev/null
+++ b/examples/code-llama/13b/lora.yml
@@ -0,0 +1,67 @@
+base_model: codellama/CodeLlama-13b-hf
+model_type: LlamaForCausalLM
+tokenizer_type: CodeLlamaTokenizer
+
+load_in_8bit: true
+load_in_4bit: false
+strict: false
+
+datasets:
+ - path: mhenrichsen/alpaca_2k_test
+ type: alpaca
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/lora-out
+
+sequence_len: 4096
+sample_packing: true
+pad_to_sequence_len: true
+
+adapter: lora
+lora_model_dir:
+lora_r: 32
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_linear: true
+lora_fan_in_fan_out:
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 2
+num_epochs: 4
+optimizer: adamw_bnb_8bit
+lr_scheduler: cosine
+learning_rate: 0.0002
+
+train_on_inputs: false
+group_by_length: false
+bf16: auto
+fp16:
+tf32: false
+
+gradient_checkpointing: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention:
+flash_attention: true
+s2_attention:
+
+warmup_steps: 10
+evals_per_epoch: 4
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:
+special_tokens:
+ bos_token: "| \s*## Table of Contents.*? | \s*
](https://github.com/OpenAccess-AI-Collective/axolotl)"""
+ transformers.modelcard.AUTOGENERATED_TRAINER_COMMENT += f"\n{badge_markdown}"
+
+ if getattr(cfg, "axolotl_config_path"):
+ raw_axolotl_cfg = Path(cfg.axolotl_config_path)
+ version = get_distribution("axolotl").version
+ if raw_axolotl_cfg.is_file():
+ transformers.modelcard.AUTOGENERATED_TRAINER_COMMENT += f"\n
+ 
+
+
+
+### Finetune Llama 3, Mistral & Gemma 2-5x faster with 80% less memory!
+
+
+
+
**Twitter (aka X)** | [Follow us on X](https://twitter.com/unslothai)|
+| 📜 **Documentation** | [Read The Doc](https://github.com/unslothai/unsloth/tree/main#-documentation) |
+| 💾 **Installation** | [unsloth/README.md](https://github.com/unslothai/unsloth/tree/main#installation-instructions)|
+| 🥇 **Benchmarking** | [Performance Tables](https://github.com/unslothai/unsloth/tree/main#-performance-benchmarking)
+| 🌐 **Released Models** | [Unsloth Releases](https://huggingface.co/unsloth)|
+| ✍️ **Blog** | [Read our Blogs](https://unsloth.ai/blog)|
+
+## ⭐ Key Features
+- All kernels written in [OpenAI's Triton](https://openai.com/research/triton) language. **Manual backprop engine**.
+- **0% loss in accuracy** - no approximation methods - all exact.
+- No change of hardware. Supports NVIDIA GPUs since 2018+. Minimum CUDA Capability 7.0 (V100, T4, Titan V, RTX 20, 30, 40x, A100, H100, L40 etc) [Check your GPU!](https://developer.nvidia.com/cuda-gpus) GTX 1070, 1080 works, but is slow.
+- Works on **Linux** and **Windows** via WSL.
+- Supports 4bit and 16bit QLoRA / LoRA finetuning via [bitsandbytes](https://github.com/TimDettmers/bitsandbytes).
+- Open source trains 5x faster - see [Unsloth Pro](https://unsloth.ai/) for up to **30x faster training**!
+- If you trained a model with 🦥Unsloth, you can use this cool sticker! 
+
+
+## 🥇 Performance Benchmarking
+- For the full list of **reproducable** benchmarking tables, [go to our website](https://unsloth.ai/blog/mistral-benchmark#Benchmark%20tables)
+
+| 1 A100 40GB | 🤗Hugging Face | Flash Attention | 🦥Unsloth Open Source | 🦥[Unsloth Pro](https://unsloth.ai/pricing) |
+|--------------|--------------|-----------------|---------------------|-----------------|
+| Alpaca | 1x | 1.04x | 1.98x | **15.64x** |
+| LAION Chip2 | 1x | 0.92x | 1.61x | **20.73x** |
+| OASST | 1x | 1.19x | 2.17x | **14.83x** |
+| Slim Orca | 1x | 1.18x | 2.22x | **14.82x** |
+
+- Benchmarking table below was conducted by [🤗Hugging Face](https://huggingface.co/blog/unsloth-trl).
+
+| Free Colab T4 | Dataset | 🤗Hugging Face | Pytorch 2.1.1 | 🦥Unsloth | 🦥 VRAM reduction |
+| --- | --- | --- | --- | --- | --- |
+| Llama-2 7b | OASST | 1x | 1.19x | 1.95x | -43.3% |
+| Mistral 7b | Alpaca | 1x | 1.07x | 1.56x | -13.7% |
+| Tiny Llama 1.1b | Alpaca | 1x | 2.06x | 3.87x | -73.8% |
+| DPO with Zephyr | Ultra Chat | 1x | 1.09x | 1.55x | -18.6% |
+
+
+
+## 💾 Installation Instructions
+### Conda Installation
+Select either `pytorch-cuda=11.8` for CUDA 11.8 or `pytorch-cuda=12.1` for CUDA 12.1. If you have `mamba`, use `mamba` instead of `conda` for faster solving. See this [Github issue](https://github.com/unslothai/unsloth/issues/73) for help on debugging Conda installs.
+```bash
+conda create --name unsloth_env python=3.10
+conda activate unsloth_env
+
+conda install pytorch-cuda=<12.1/11.8> pytorch cudatoolkit xformers -c pytorch -c nvidia -c xformers
+
+pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
+
+pip install --no-deps trl peft accelerate bitsandbytes
+```
+
+### Pip Installation
+Do **NOT** use this if you have Anaconda. You must use the Conda install method, or else stuff will BREAK.
+
+1. Find your CUDA version via
+```python
+import torch; torch.version.cuda
+```
+2. For Pytorch 2.1.0: You can update Pytorch via Pip (interchange `cu121` / `cu118`). Go to https://pytorch.org/ to learn more. Select either `cu118` for CUDA 11.8 or `cu121` for CUDA 12.1. If you have a RTX 3060 or higher (A100, H100 etc), use the `"ampere"` path. For Pytorch 2.1.1: go to step 3. For Pytorch 2.2.0: go to step 4.
+```bash
+pip install --upgrade --force-reinstall --no-cache-dir torch==2.1.0 triton \
+ --index-url https://download.pytorch.org/whl/cu121
+```
+```bash
+pip install "unsloth[cu118] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu118-ampere] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-ampere] @ git+https://github.com/unslothai/unsloth.git"
+```
+3. For Pytorch 2.1.1: Use the `"ampere"` path for newer RTX 30xx GPUs or higher.
+```bash
+pip install --upgrade --force-reinstall --no-cache-dir torch==2.1.1 triton \
+ --index-url https://download.pytorch.org/whl/cu121
+```
+```bash
+pip install "unsloth[cu118-torch211] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-torch211] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu118-ampere-torch211] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-ampere-torch211] @ git+https://github.com/unslothai/unsloth.git"
+```
+4. For Pytorch 2.2.0: Use the `"ampere"` path for newer RTX 30xx GPUs or higher.
+```bash
+pip install --upgrade --force-reinstall --no-cache-dir torch==2.2.0 triton \
+ --index-url https://download.pytorch.org/whl/cu121
+```
+```bash
+pip install "unsloth[cu118-torch220] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-torch220] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu118-ampere-torch220] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-ampere-torch220] @ git+https://github.com/unslothai/unsloth.git"
+```
+5. If you get errors, try the below first, then go back to step 1:
+```bash
+pip install --upgrade pip
+```
+6. For Pytorch 2.2.1:
+```bash
+# RTX 3090, 4090 Ampere GPUs:
+pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
+pip install --no-deps packaging ninja einops flash-attn xformers trl peft accelerate bitsandbytes
+
+# Pre Ampere RTX 2080, T4, GTX 1080 GPUs:
+pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
+pip install --no-deps xformers trl peft accelerate bitsandbytes
+```
+7. For Pytorch 2.3.0: Use the `"ampere"` path for newer RTX 30xx GPUs or higher.
+```bash
+pip install "unsloth[cu118-torch230] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-torch230] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu118-ampere-torch230] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-ampere-torch230] @ git+https://github.com/unslothai/unsloth.git"
+```
+8. To troubleshoot installs try the below (all must succeed). Xformers should mostly all be available.
+```bash
+nvcc
+python -m xformers.info
+python -m bitsandbytes
+```
+
+## 📜 Documentation
+- Go to our [Wiki page](https://github.com/unslothai/unsloth/wiki) for saving to GGUF, checkpointing, evaluation and more!
+- We support Huggingface's TRL, Trainer, Seq2SeqTrainer or even Pytorch code!
+- We're in 🤗Hugging Face's official docs! Check out the [SFT docs](https://huggingface.co/docs/trl/main/en/sft_trainer#accelerate-fine-tuning-2x-using-unsloth) and [DPO docs](https://huggingface.co/docs/trl/main/en/dpo_trainer#accelerate-dpo-fine-tuning-using-unsloth)!
+
+```python
+from unsloth import FastLanguageModel
+import torch
+from trl import SFTTrainer
+from transformers import TrainingArguments
+from datasets import load_dataset
+max_seq_length = 2048 # Supports RoPE Scaling interally, so choose any!
+# Get LAION dataset
+url = "https://huggingface.co/datasets/laion/OIG/resolve/main/unified_chip2.jsonl"
+dataset = load_dataset("json", data_files = {"train" : url}, split = "train")
+
+# 4bit pre quantized models we support for 4x faster downloading + no OOMs.
+fourbit_models = [
+ "unsloth/mistral-7b-bnb-4bit",
+ "unsloth/mistral-7b-instruct-v0.2-bnb-4bit",
+ "unsloth/llama-2-7b-bnb-4bit",
+ "unsloth/gemma-7b-bnb-4bit",
+ "unsloth/gemma-7b-it-bnb-4bit", # Instruct version of Gemma 7b
+ "unsloth/gemma-2b-bnb-4bit",
+ "unsloth/gemma-2b-it-bnb-4bit", # Instruct version of Gemma 2b
+ "unsloth/llama-3-8b-bnb-4bit", # [NEW] 15 Trillion token Llama-3
+ "unsloth/Phi-3-mini-4k-instruct-bnb-4bit",
+] # More models at https://huggingface.co/unsloth
+
+model, tokenizer = FastLanguageModel.from_pretrained(
+ model_name = "unsloth/llama-3-8b-bnb-4bit",
+ max_seq_length = max_seq_length,
+ dtype = None,
+ load_in_4bit = True,
+)
+
+# Do model patching and add fast LoRA weights
+model = FastLanguageModel.get_peft_model(
+ model,
+ r = 16,
+ target_modules = ["q_proj", "k_proj", "v_proj", "o_proj",
+ "gate_proj", "up_proj", "down_proj",],
+ lora_alpha = 16,
+ 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,
+ max_seq_length = max_seq_length,
+ use_rslora = False, # We support rank stabilized LoRA
+ loftq_config = None, # And LoftQ
+)
+
+trainer = SFTTrainer(
+ model = model,
+ train_dataset = dataset,
+ dataset_text_field = "text",
+ max_seq_length = max_seq_length,
+ tokenizer = tokenizer,
+ args = TrainingArguments(
+ per_device_train_batch_size = 2,
+ gradient_accumulation_steps = 4,
+ warmup_steps = 10,
+ max_steps = 60,
+ fp16 = not torch.cuda.is_bf16_supported(),
+ bf16 = torch.cuda.is_bf16_supported(),
+ logging_steps = 1,
+ output_dir = "outputs",
+ optim = "adamw_8bit",
+ seed = 3407,
+ ),
+)
+trainer.train()
+
+# Go to https://github.com/unslothai/unsloth/wiki for advanced tips like
+# (1) Saving to GGUF / merging to 16bit for vLLM
+# (2) Continued training from a saved LoRA adapter
+# (3) Adding an evaluation loop / OOMs
+# (4) Cutomized chat templates
+```
+
+
+## DPO Support
+DPO (Direct Preference Optimization), PPO, Reward Modelling all seem to work as per 3rd party independent testing from [Llama-Factory](https://github.com/hiyouga/LLaMA-Factory). We have a preliminary Google Colab notebook for reproducing Zephyr on Tesla T4 here: [notebook](https://colab.research.google.com/drive/15vttTpzzVXv_tJwEk-hIcQ0S9FcEWvwP?usp=sharing).
+
+We're in 🤗Hugging Face's official docs! We're on the [SFT docs](https://huggingface.co/docs/trl/main/en/sft_trainer#accelerate-fine-tuning-2x-using-unsloth) and the [DPO docs](https://huggingface.co/docs/trl/main/en/dpo_trainer#accelerate-dpo-fine-tuning-using-unsloth)!
+
+```python
+from unsloth import FastLanguageModel, PatchDPOTrainer
+PatchDPOTrainer()
+import torch
+from transformers import TrainingArguments
+from trl import DPOTrainer
+
+model, tokenizer = FastLanguageModel.from_pretrained(
+ model_name = "unsloth/zephyr-sft-bnb-4bit",
+ max_seq_length = max_seq_length,
+ dtype = None,
+ load_in_4bit = True,
+)
+
+# Do model patching and add fast LoRA weights
+model = FastLanguageModel.get_peft_model(
+ model,
+ r = 64,
+ 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,
+ max_seq_length = max_seq_length,
+)
+
+dpo_trainer = DPOTrainer(
+ model = model,
+ ref_model = None,
+ args = TrainingArguments(
+ per_device_train_batch_size = 4,
+ gradient_accumulation_steps = 8,
+ warmup_ratio = 0.1,
+ num_train_epochs = 3,
+ fp16 = not torch.cuda.is_bf16_supported(),
+ bf16 = torch.cuda.is_bf16_supported(),
+ logging_steps = 1,
+ optim = "adamw_8bit",
+ seed = 42,
+ output_dir = "outputs",
+ ),
+ beta = 0.1,
+ train_dataset = YOUR_DATASET_HERE,
+ # eval_dataset = YOUR_DATASET_HERE,
+ tokenizer = tokenizer,
+ max_length = 1024,
+ max_prompt_length = 512,
+)
+dpo_trainer.train()
+```
+
+## 🥇 Detailed Benchmarking Tables
+- Click "Code" for fully reproducible examples
+- "Unsloth Equal" is a preview of our PRO version, with code stripped out. All settings and the loss curve remains identical.
+- For the full list of benchmarking tables, [go to our website](https://unsloth.ai/blog/mistral-benchmark#Benchmark%20tables)
+
+| 1 A100 40GB | 🤗Hugging Face | Flash Attention 2 | 🦥Unsloth Open | Unsloth Equal | Unsloth Pro | Unsloth Max |
+|--------------|-------------|-------------|-----------------|--------------|---------------|-------------|
+| Alpaca | 1x | 1.04x | 1.98x | 2.48x | 5.32x | **15.64x** |
+| code | [Code](https://colab.research.google.com/drive/1u4dBeM-0vGNVmmO6X7cScAut-Hyt4KDF?usp=sharing) | [Code](https://colab.research.google.com/drive/1fgTOxpMbVjloQBvZyz4lF4BacKSZOB2A?usp=sharing) | [Code](https://colab.research.google.com/drive/1YIPY_18xm-K0iJDgvNkRoJsgkPMPAO3G?usp=sharing) | [Code](https://colab.research.google.com/drive/1ANW8EFL3LVyTD7Gq4TkheC1Z7Rxw-rHp?usp=sharing) | | |
+| seconds| 1040 | 1001 | 525 | 419 | 196 | 67 |
+| memory MB| 18235 | 15365 | 9631 | 8525 | | |
+| % saved| | 15.74 | 47.18 | 53.25 | | | |
+
+### Llama-Factory 3rd party benchmarking
+- [Link to performance table.](https://github.com/hiyouga/LLaMA-Factory/wiki/Performance-Comparison) TGS: tokens per GPU per second. Model: LLaMA2-7B. GPU: NVIDIA A100 * 1. Batch size: 4. Gradient accumulation: 2. LoRA rank: 8. Max length: 1024.
+
+| Method | Bits | TGS | GRAM | Speed |
+| --- | --- | --- | --- | --- |
+| HF | 16 | 2392 | 18GB | 100% |
+| HF+FA2 | 16 | 2954 | 17GB | 123% |
+| Unsloth+FA2 | 16 | 4007 | 16GB | **168%** |
+| HF | 4 | 2415 | 9GB | 101% |
+| Unsloth+FA2 | 4 | 3726 | 7GB | **160%** |
+
+### Performance comparisons between popular models
+
+
+
+
+
+### Finetune Llama 3, Mistral & Gemma 2-5x faster with 80% less memory!
+
+
+
+ **Twitter (aka X)** | [Follow us on X](https://twitter.com/unslothai)|
+| 📜 **Documentation** | [Read The Doc](https://github.com/unslothai/unsloth/tree/main#-documentation) |
+| 💾 **Installation** | [unsloth/README.md](https://github.com/unslothai/unsloth/tree/main#installation-instructions)|
+| 🥇 **Benchmarking** | [Performance Tables](https://github.com/unslothai/unsloth/tree/main#-performance-benchmarking)
+| 🌐 **Released Models** | [Unsloth Releases](https://huggingface.co/unsloth)|
+| ✍️ **Blog** | [Read our Blogs](https://unsloth.ai/blog)|
+
+## ⭐ Key Features
+- All kernels written in [OpenAI's Triton](https://openai.com/research/triton) language. **Manual backprop engine**.
+- **0% loss in accuracy** - no approximation methods - all exact.
+- No change of hardware. Supports NVIDIA GPUs since 2018+. Minimum CUDA Capability 7.0 (V100, T4, Titan V, RTX 20, 30, 40x, A100, H100, L40 etc) [Check your GPU!](https://developer.nvidia.com/cuda-gpus) GTX 1070, 1080 works, but is slow.
+- Works on **Linux** and **Windows** via WSL.
+- Supports 4bit and 16bit QLoRA / LoRA finetuning via [bitsandbytes](https://github.com/TimDettmers/bitsandbytes).
+- Open source trains 5x faster - see [Unsloth Pro](https://unsloth.ai/) for up to **30x faster training**!
+- If you trained a model with 🦥Unsloth, you can use this cool sticker!
+
+
+## 🥇 Performance Benchmarking
+- For the full list of **reproducable** benchmarking tables, [go to our website](https://unsloth.ai/blog/mistral-benchmark#Benchmark%20tables)
+
+| 1 A100 40GB | 🤗Hugging Face | Flash Attention | 🦥Unsloth Open Source | 🦥[Unsloth Pro](https://unsloth.ai/pricing) |
+|--------------|--------------|-----------------|---------------------|-----------------|
+| Alpaca | 1x | 1.04x | 1.98x | **15.64x** |
+| LAION Chip2 | 1x | 0.92x | 1.61x | **20.73x** |
+| OASST | 1x | 1.19x | 2.17x | **14.83x** |
+| Slim Orca | 1x | 1.18x | 2.22x | **14.82x** |
+
+- Benchmarking table below was conducted by [🤗Hugging Face](https://huggingface.co/blog/unsloth-trl).
+
+| Free Colab T4 | Dataset | 🤗Hugging Face | Pytorch 2.1.1 | 🦥Unsloth | 🦥 VRAM reduction |
+| --- | --- | --- | --- | --- | --- |
+| Llama-2 7b | OASST | 1x | 1.19x | 1.95x | -43.3% |
+| Mistral 7b | Alpaca | 1x | 1.07x | 1.56x | -13.7% |
+| Tiny Llama 1.1b | Alpaca | 1x | 2.06x | 3.87x | -73.8% |
+| DPO with Zephyr | Ultra Chat | 1x | 1.09x | 1.55x | -18.6% |
+
+
+
+## 💾 Installation Instructions
+### Conda Installation
+Select either `pytorch-cuda=11.8` for CUDA 11.8 or `pytorch-cuda=12.1` for CUDA 12.1. If you have `mamba`, use `mamba` instead of `conda` for faster solving. See this [Github issue](https://github.com/unslothai/unsloth/issues/73) for help on debugging Conda installs.
+```bash
+conda create --name unsloth_env python=3.10
+conda activate unsloth_env
+
+conda install pytorch-cuda=<12.1/11.8> pytorch cudatoolkit xformers -c pytorch -c nvidia -c xformers
+
+pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
+
+pip install --no-deps trl peft accelerate bitsandbytes
+```
+
+### Pip Installation
+Do **NOT** use this if you have Anaconda. You must use the Conda install method, or else stuff will BREAK.
+
+1. Find your CUDA version via
+```python
+import torch; torch.version.cuda
+```
+2. For Pytorch 2.1.0: You can update Pytorch via Pip (interchange `cu121` / `cu118`). Go to https://pytorch.org/ to learn more. Select either `cu118` for CUDA 11.8 or `cu121` for CUDA 12.1. If you have a RTX 3060 or higher (A100, H100 etc), use the `"ampere"` path. For Pytorch 2.1.1: go to step 3. For Pytorch 2.2.0: go to step 4.
+```bash
+pip install --upgrade --force-reinstall --no-cache-dir torch==2.1.0 triton \
+ --index-url https://download.pytorch.org/whl/cu121
+```
+```bash
+pip install "unsloth[cu118] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu118-ampere] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-ampere] @ git+https://github.com/unslothai/unsloth.git"
+```
+3. For Pytorch 2.1.1: Use the `"ampere"` path for newer RTX 30xx GPUs or higher.
+```bash
+pip install --upgrade --force-reinstall --no-cache-dir torch==2.1.1 triton \
+ --index-url https://download.pytorch.org/whl/cu121
+```
+```bash
+pip install "unsloth[cu118-torch211] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-torch211] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu118-ampere-torch211] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-ampere-torch211] @ git+https://github.com/unslothai/unsloth.git"
+```
+4. For Pytorch 2.2.0: Use the `"ampere"` path for newer RTX 30xx GPUs or higher.
+```bash
+pip install --upgrade --force-reinstall --no-cache-dir torch==2.2.0 triton \
+ --index-url https://download.pytorch.org/whl/cu121
+```
+```bash
+pip install "unsloth[cu118-torch220] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-torch220] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu118-ampere-torch220] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-ampere-torch220] @ git+https://github.com/unslothai/unsloth.git"
+```
+5. If you get errors, try the below first, then go back to step 1:
+```bash
+pip install --upgrade pip
+```
+6. For Pytorch 2.2.1:
+```bash
+# RTX 3090, 4090 Ampere GPUs:
+pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
+pip install --no-deps packaging ninja einops flash-attn xformers trl peft accelerate bitsandbytes
+
+# Pre Ampere RTX 2080, T4, GTX 1080 GPUs:
+pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
+pip install --no-deps xformers trl peft accelerate bitsandbytes
+```
+7. For Pytorch 2.3.0: Use the `"ampere"` path for newer RTX 30xx GPUs or higher.
+```bash
+pip install "unsloth[cu118-torch230] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-torch230] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu118-ampere-torch230] @ git+https://github.com/unslothai/unsloth.git"
+pip install "unsloth[cu121-ampere-torch230] @ git+https://github.com/unslothai/unsloth.git"
+```
+8. To troubleshoot installs try the below (all must succeed). Xformers should mostly all be available.
+```bash
+nvcc
+python -m xformers.info
+python -m bitsandbytes
+```
+
+## 📜 Documentation
+- Go to our [Wiki page](https://github.com/unslothai/unsloth/wiki) for saving to GGUF, checkpointing, evaluation and more!
+- We support Huggingface's TRL, Trainer, Seq2SeqTrainer or even Pytorch code!
+- We're in 🤗Hugging Face's official docs! Check out the [SFT docs](https://huggingface.co/docs/trl/main/en/sft_trainer#accelerate-fine-tuning-2x-using-unsloth) and [DPO docs](https://huggingface.co/docs/trl/main/en/dpo_trainer#accelerate-dpo-fine-tuning-using-unsloth)!
+
+```python
+from unsloth import FastLanguageModel
+import torch
+from trl import SFTTrainer
+from transformers import TrainingArguments
+from datasets import load_dataset
+max_seq_length = 2048 # Supports RoPE Scaling interally, so choose any!
+# Get LAION dataset
+url = "https://huggingface.co/datasets/laion/OIG/resolve/main/unified_chip2.jsonl"
+dataset = load_dataset("json", data_files = {"train" : url}, split = "train")
+
+# 4bit pre quantized models we support for 4x faster downloading + no OOMs.
+fourbit_models = [
+ "unsloth/mistral-7b-bnb-4bit",
+ "unsloth/mistral-7b-instruct-v0.2-bnb-4bit",
+ "unsloth/llama-2-7b-bnb-4bit",
+ "unsloth/gemma-7b-bnb-4bit",
+ "unsloth/gemma-7b-it-bnb-4bit", # Instruct version of Gemma 7b
+ "unsloth/gemma-2b-bnb-4bit",
+ "unsloth/gemma-2b-it-bnb-4bit", # Instruct version of Gemma 2b
+ "unsloth/llama-3-8b-bnb-4bit", # [NEW] 15 Trillion token Llama-3
+ "unsloth/Phi-3-mini-4k-instruct-bnb-4bit",
+] # More models at https://huggingface.co/unsloth
+
+model, tokenizer = FastLanguageModel.from_pretrained(
+ model_name = "unsloth/llama-3-8b-bnb-4bit",
+ max_seq_length = max_seq_length,
+ dtype = None,
+ load_in_4bit = True,
+)
+
+# Do model patching and add fast LoRA weights
+model = FastLanguageModel.get_peft_model(
+ model,
+ r = 16,
+ target_modules = ["q_proj", "k_proj", "v_proj", "o_proj",
+ "gate_proj", "up_proj", "down_proj",],
+ lora_alpha = 16,
+ 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,
+ max_seq_length = max_seq_length,
+ use_rslora = False, # We support rank stabilized LoRA
+ loftq_config = None, # And LoftQ
+)
+
+trainer = SFTTrainer(
+ model = model,
+ train_dataset = dataset,
+ dataset_text_field = "text",
+ max_seq_length = max_seq_length,
+ tokenizer = tokenizer,
+ args = TrainingArguments(
+ per_device_train_batch_size = 2,
+ gradient_accumulation_steps = 4,
+ warmup_steps = 10,
+ max_steps = 60,
+ fp16 = not torch.cuda.is_bf16_supported(),
+ bf16 = torch.cuda.is_bf16_supported(),
+ logging_steps = 1,
+ output_dir = "outputs",
+ optim = "adamw_8bit",
+ seed = 3407,
+ ),
+)
+trainer.train()
+
+# Go to https://github.com/unslothai/unsloth/wiki for advanced tips like
+# (1) Saving to GGUF / merging to 16bit for vLLM
+# (2) Continued training from a saved LoRA adapter
+# (3) Adding an evaluation loop / OOMs
+# (4) Cutomized chat templates
+```
+
+
+## DPO Support
+DPO (Direct Preference Optimization), PPO, Reward Modelling all seem to work as per 3rd party independent testing from [Llama-Factory](https://github.com/hiyouga/LLaMA-Factory). We have a preliminary Google Colab notebook for reproducing Zephyr on Tesla T4 here: [notebook](https://colab.research.google.com/drive/15vttTpzzVXv_tJwEk-hIcQ0S9FcEWvwP?usp=sharing).
+
+We're in 🤗Hugging Face's official docs! We're on the [SFT docs](https://huggingface.co/docs/trl/main/en/sft_trainer#accelerate-fine-tuning-2x-using-unsloth) and the [DPO docs](https://huggingface.co/docs/trl/main/en/dpo_trainer#accelerate-dpo-fine-tuning-using-unsloth)!
+
+```python
+from unsloth import FastLanguageModel, PatchDPOTrainer
+PatchDPOTrainer()
+import torch
+from transformers import TrainingArguments
+from trl import DPOTrainer
+
+model, tokenizer = FastLanguageModel.from_pretrained(
+ model_name = "unsloth/zephyr-sft-bnb-4bit",
+ max_seq_length = max_seq_length,
+ dtype = None,
+ load_in_4bit = True,
+)
+
+# Do model patching and add fast LoRA weights
+model = FastLanguageModel.get_peft_model(
+ model,
+ r = 64,
+ 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,
+ max_seq_length = max_seq_length,
+)
+
+dpo_trainer = DPOTrainer(
+ model = model,
+ ref_model = None,
+ args = TrainingArguments(
+ per_device_train_batch_size = 4,
+ gradient_accumulation_steps = 8,
+ warmup_ratio = 0.1,
+ num_train_epochs = 3,
+ fp16 = not torch.cuda.is_bf16_supported(),
+ bf16 = torch.cuda.is_bf16_supported(),
+ logging_steps = 1,
+ optim = "adamw_8bit",
+ seed = 42,
+ output_dir = "outputs",
+ ),
+ beta = 0.1,
+ train_dataset = YOUR_DATASET_HERE,
+ # eval_dataset = YOUR_DATASET_HERE,
+ tokenizer = tokenizer,
+ max_length = 1024,
+ max_prompt_length = 512,
+)
+dpo_trainer.train()
+```
+
+## 🥇 Detailed Benchmarking Tables
+- Click "Code" for fully reproducible examples
+- "Unsloth Equal" is a preview of our PRO version, with code stripped out. All settings and the loss curve remains identical.
+- For the full list of benchmarking tables, [go to our website](https://unsloth.ai/blog/mistral-benchmark#Benchmark%20tables)
+
+| 1 A100 40GB | 🤗Hugging Face | Flash Attention 2 | 🦥Unsloth Open | Unsloth Equal | Unsloth Pro | Unsloth Max |
+|--------------|-------------|-------------|-----------------|--------------|---------------|-------------|
+| Alpaca | 1x | 1.04x | 1.98x | 2.48x | 5.32x | **15.64x** |
+| code | [Code](https://colab.research.google.com/drive/1u4dBeM-0vGNVmmO6X7cScAut-Hyt4KDF?usp=sharing) | [Code](https://colab.research.google.com/drive/1fgTOxpMbVjloQBvZyz4lF4BacKSZOB2A?usp=sharing) | [Code](https://colab.research.google.com/drive/1YIPY_18xm-K0iJDgvNkRoJsgkPMPAO3G?usp=sharing) | [Code](https://colab.research.google.com/drive/1ANW8EFL3LVyTD7Gq4TkheC1Z7Rxw-rHp?usp=sharing) | | |
+| seconds| 1040 | 1001 | 525 | 419 | 196 | 67 |
+| memory MB| 18235 | 15365 | 9631 | 8525 | | |
+| % saved| | 15.74 | 47.18 | 53.25 | | | |
+
+### Llama-Factory 3rd party benchmarking
+- [Link to performance table.](https://github.com/hiyouga/LLaMA-Factory/wiki/Performance-Comparison) TGS: tokens per GPU per second. Model: LLaMA2-7B. GPU: NVIDIA A100 * 1. Batch size: 4. Gradient accumulation: 2. LoRA rank: 8. Max length: 1024.
+
+| Method | Bits | TGS | GRAM | Speed |
+| --- | --- | --- | --- | --- |
+| HF | 16 | 2392 | 18GB | 100% |
+| HF+FA2 | 16 | 2954 | 17GB | 123% |
+| Unsloth+FA2 | 16 | 4007 | 16GB | **168%** |
+| HF | 4 | 2415 | 9GB | 101% |
+| Unsloth+FA2 | 4 | 3726 | 7GB | **160%** |
+
+### Performance comparisons between popular models
+
](https://github.com/unslothai/unsloth)
+"""
+
+
+def _determine_username(save_directory, old_username, token):
+ username = ""
+ save_directory = save_directory.lstrip("./")
+ if "/" not in save_directory:
+ from huggingface_hub import whoami
+ try:
+ username = whoami(token = token)["name"]
+ if type(old_username) is str and username != old_username:
+ username = old_username
+ pass
+ save_directory = f"{username}/{save_directory}"
+ except:
+ raise RuntimeError(f"Unsloth: {save_directory} is not a Huggingface directory.")
+ else:
+ username = save_directory.split("/")[0]
+ pass
+ return save_directory, username
+pass
+
+
+def upload_to_huggingface(
+ model,
+ save_directory,
+ token,
+ method,
+ extra = "",
+ file_location = None,
+ old_username = None,
+ private = None,
+):
+ save_directory, username = _determine_username(save_directory, old_username, token)
+
+ from huggingface_hub import create_repo
+ try:
+ create_repo(
+ repo_id = save_directory,
+ token = token,
+ repo_type = "model",
+ exist_ok = False,
+ private = private,
+ )
+
+ # Create model card
+ from huggingface_hub import ModelCard
+ content = MODEL_CARD.format(
+ username = username,
+ base_model = model.config._name_or_path,
+ model_type = model.config.model_type,
+ method = "",
+ extra = extra,
+ )
+ card = ModelCard(content)
+ card.push_to_hub(save_directory, token = token)
+ except:
+ pass
+
+ if file_location is not None:
+ # Now upload file
+ from huggingface_hub import HfApi
+ hf_api = HfApi(token = token)
+
+ if "/" in file_location:
+ uploaded_location = file_location[file_location.rfind("/")+1:]
+ else:
+ uploaded_location = file_location
+ pass
+
+ hf_api.upload_file(
+ path_or_fileobj = file_location,
+ path_in_repo = uploaded_location,
+ repo_id = save_directory,
+ repo_type = "model",
+ commit_message = "(Trained with Unsloth)",
+ )
+
+ # We also upload a config.json file
+ import json
+ with open("_temporary_unsloth_config.json", "w") as file:
+ json.dump({"model_type" : model.config.model_type}, file, indent = 4)
+ pass
+ hf_api.upload_file(
+ path_or_fileobj = "_temporary_unsloth_config.json",
+ path_in_repo = "config.json",
+ repo_id = save_directory,
+ repo_type = "model",
+ commit_message = "(Trained with Unsloth)",
+ )
+ os.remove("_temporary_unsloth_config.json")
+ pass
+ return username
+pass
+
+
+def unsloth_save_pretrained_gguf(
+ self,
+ save_directory : Union[str, os.PathLike],
+ tokenizer = None,
+ quantization_method : str = "fast_quantized",
+ first_conversion : str = "f16",
+ push_to_hub : bool = False,
+ token : Optional[Union[str, bool]] = None,
+ private : Optional[bool] = None,
+ is_main_process : bool = True,
+ state_dict : Optional[dict] = None,
+ save_function : Callable = torch.save,
+ max_shard_size : Union[int, str] = "5GB",
+ safe_serialization : bool = True,
+ variant : Optional[str] = None,
+ save_peft_format : bool = True,
+ tags : List[str] = None,
+ temporary_location : str = "_unsloth_temporary_saved_buffers",
+ maximum_memory_usage : float = 0.85,
+):
+ """
+ Same as .save_pretrained(...) except 4bit weights are auto
+ converted to float16 then converted to GGUF / llama.cpp format.
+
+ Choose for `quantization_method` to be:
+ "not_quantized" : "Recommended. Fast conversion. Slow inference, big files.",
+ "fast_quantized" : "Recommended. Fast conversion. OK inference, OK file size.",
+ "quantized" : "Recommended. Slow conversion. Fast inference, small files.",
+ "f32" : "Not recommended. Retains 100% accuracy, but super slow and memory hungry.",
+ "f16" : "Fastest conversion + retains 100% accuracy. Slow and memory hungry.",
+ "q8_0" : "Fast conversion. High resource use, but generally acceptable.",
+ "q4_k_m" : "Recommended. Uses Q6_K for half of the attention.wv and feed_forward.w2 tensors, else Q4_K",
+ "q5_k_m" : "Recommended. Uses Q6_K for half of the attention.wv and feed_forward.w2 tensors, else Q5_K",
+ "q2_k" : "Uses Q4_K for the attention.vw and feed_forward.w2 tensors, Q2_K for the other tensors.",
+ "q3_k_l" : "Uses Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else Q3_K",
+ "q3_k_m" : "Uses Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else Q3_K",
+ "q3_k_s" : "Uses Q3_K for all tensors",
+ "q4_0" : "Original quant method, 4-bit.",
+ "q4_1" : "Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models.",
+ "q4_k_s" : "Uses Q4_K for all tensors",
+ "q4_k" : "alias for q4_k_m",
+ "q5_k" : "alias for q5_k_m",
+ "q5_0" : "Higher accuracy, higher resource usage and slower inference.",
+ "q5_1" : "Even higher accuracy, resource usage and slower inference.",
+ "q5_k_s" : "Uses Q5_K for all tensors",
+ "q6_k" : "Uses Q8_K for all tensors",
+ "iq2_xxs" : "2.06 bpw quantization",
+ "iq2_xs" : "2.31 bpw quantization",
+ "iq3_xxs" : "3.06 bpw quantization",
+ "q3_k_xs" : "3-bit extra small quantization",
+ """
+ if tokenizer is None:
+ raise ValueError("Unsloth: Saving to GGUF must have a tokenizer.")
+
+ arguments = dict(locals())
+ arguments["model"] = self
+ arguments["tokenizer"] = tokenizer
+ arguments["push_to_hub"] = False # We save ourselves
+ arguments["save_method"] = "merged_16bit" # Must be 16bit
+ del arguments["self"]
+ del arguments["quantization_method"]
+ del arguments["first_conversion"]
+
+ # Non blocking install GGUF first
+ if not os.path.exists("llama.cpp"):
+
+ if IS_KAGGLE_ENVIRONMENT:
+ # Kaggle is weird - no blocking installs, and no CUDA?
+ python_install = install_python_non_blocking(["gguf", "protobuf"])
+ python_install.wait()
+ install_llama_cpp_blocking(use_cuda = False)
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ makefile = None
+ else:
+ git_clone = install_llama_cpp_clone_non_blocking()
+ python_install = install_python_non_blocking(["gguf", "protobuf"])
+ git_clone.wait()
+ makefile = install_llama_cpp_make_non_blocking()
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ python_install.wait()
+ pass
+ else:
+ try:
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ makefile = None
+ except:
+ # Retry by recloning llama.cpp
+ if IS_KAGGLE_ENVIRONMENT:
+ # Kaggle is weird - no blocking installs, and no CUDA?
+ python_install = install_python_non_blocking(["gguf", "protobuf"])
+ python_install.wait()
+ install_llama_cpp_blocking(use_cuda = False)
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ makefile = None
+ else:
+ git_clone = install_llama_cpp_clone_non_blocking()
+ python_install = install_python_non_blocking(["gguf", "protobuf"])
+ git_clone.wait()
+ makefile = install_llama_cpp_make_non_blocking()
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ python_install.wait()
+ pass
+ pass
+ pass
+
+ for _ in range(3):
+ gc.collect()
+
+ model_type = self.config.model_type
+ is_sentencepiece_model = check_if_sentencepiece_model(self)
+ file_location = save_to_gguf(model_type, is_sentencepiece_model,
+ new_save_directory, quantization_method, first_conversion, makefile,
+ )
+
+ if push_to_hub:
+ print("Unsloth: Uploading GGUF to Huggingface Hub...")
+ username = upload_to_huggingface(
+ self, save_directory, token,
+ "GGUF converted", "gguf", file_location, old_username, private,
+ )
+ link = f"{username}/{new_save_directory.lstrip('/.')}" \
+ if username not in new_save_directory else \
+ new_save_directory.lstrip('/.')
+ print(f"Saved GGUF to https://huggingface.co/{link}")
+ pass
+pass
+
+
+def unsloth_push_to_hub_gguf(
+ self,
+ repo_id : str,
+ tokenizer = None,
+ quantization_method : str = "fast_quantized",
+ first_conversion : str = "f16",
+ use_temp_dir : Optional[bool] = None,
+ commit_message : Optional[str] = "Trained with Unsloth",
+ private : Optional[bool] = None,
+ token : Union[bool, str, None] = None,
+ max_shard_size : Union[int, str, None] = "5GB",
+ create_pr : bool = False,
+ safe_serialization : bool = True,
+ revision : str = None,
+ commit_description : str = "Upload model trained with Unsloth 2x faster",
+ tags : Optional[List[str]] = None,
+ temporary_location : str = "_unsloth_temporary_saved_buffers",
+ maximum_memory_usage : float = 0.85,
+):
+ """
+ Same as .push_to_hub(...) except 4bit weights are auto
+ converted to float16 then converted to GGUF / llama.cpp format.
+
+ Choose for `quantization_method` to be:
+ "not_quantized" : "Recommended. Fast conversion. Slow inference, big files.",
+ "fast_quantized" : "Recommended. Fast conversion. OK inference, OK file size.",
+ "quantized" : "Recommended. Slow conversion. Fast inference, small files.",
+ "f32" : "Not recommended. Retains 100% accuracy, but super slow and memory hungry.",
+ "f16" : "Fastest conversion + retains 100% accuracy. Slow and memory hungry.",
+ "q8_0" : "Fast conversion. High resource use, but generally acceptable.",
+ "q4_k_m" : "Recommended. Uses Q6_K for half of the attention.wv and feed_forward.w2 tensors, else Q4_K",
+ "q5_k_m" : "Recommended. Uses Q6_K for half of the attention.wv and feed_forward.w2 tensors, else Q5_K",
+ "q2_k" : "Uses Q4_K for the attention.vw and feed_forward.w2 tensors, Q2_K for the other tensors.",
+ "q3_k_l" : "Uses Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else Q3_K",
+ "q3_k_m" : "Uses Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else Q3_K",
+ "q3_k_s" : "Uses Q3_K for all tensors",
+ "q4_0" : "Original quant method, 4-bit.",
+ "q4_1" : "Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models.",
+ "q4_k_s" : "Uses Q4_K for all tensors",
+ "q5_0" : "Higher accuracy, higher resource usage and slower inference.",
+ "q5_1" : "Even higher accuracy, resource usage and slower inference.",
+ "q5_k_s" : "Uses Q5_K for all tensors",
+ "q6_k" : "Uses Q8_K for all tensors",
+ """
+ if tokenizer is None:
+ raise ValueError("Unsloth: Saving to GGUF must have a tokenizer.")
+
+ arguments = dict(locals())
+ arguments["model"] = self
+ arguments["tokenizer"] = tokenizer
+ arguments["save_directory"] = repo_id
+ arguments["push_to_hub"] = False # We save ourselves
+ arguments["save_method"] = "merged_16bit" # Must be 16bit
+ del arguments["self"]
+ del arguments["repo_id"]
+ del arguments["quantization_method"]
+ del arguments["first_conversion"]
+
+ # Non blocking install GGUF first
+ if not os.path.exists("llama.cpp"):
+
+ if IS_KAGGLE_ENVIRONMENT:
+ # Kaggle is weird - no blocking installs, and no CUDA?
+ python_install = install_python_non_blocking(["gguf", "protobuf"])
+ python_install.wait()
+ install_llama_cpp_blocking(use_cuda = False)
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ makefile = None
+ else:
+ git_clone = install_llama_cpp_clone_non_blocking()
+ python_install = install_python_non_blocking(["gguf", "protobuf"])
+ git_clone.wait()
+ makefile = install_llama_cpp_make_non_blocking()
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ python_install.wait()
+ pass
+ else:
+ try:
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ makefile = None
+ except:
+ # Retry by recloning llama.cpp
+ if IS_KAGGLE_ENVIRONMENT:
+ # Kaggle is weird - no blocking installs, and no CUDA?
+ python_install = install_python_non_blocking(["gguf", "protobuf"])
+ python_install.wait()
+ install_llama_cpp_blocking(use_cuda = False)
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ makefile = None
+ else:
+ git_clone = install_llama_cpp_clone_non_blocking()
+ python_install = install_python_non_blocking(["gguf", "protobuf"])
+ git_clone.wait()
+ makefile = install_llama_cpp_make_non_blocking()
+ new_save_directory, old_username = unsloth_save_model(**arguments)
+ python_install.wait()
+ pass
+ pass
+ pass
+
+ for _ in range(3):
+ gc.collect()
+
+ model_type = self.config.model_type
+ is_sentencepiece_model = check_if_sentencepiece_model(self)
+ file_location = save_to_gguf(model_type, is_sentencepiece_model,
+ new_save_directory, quantization_method, first_conversion, makefile,
+ )
+
+ print("Unsloth: Uploading GGUF to Huggingface Hub...")
+ username = upload_to_huggingface(
+ self, repo_id, token,
+ "GGUF converted", "gguf", file_location, old_username, private,
+ )
+ link = f"{username}/{new_save_directory.lstrip('/.')}" \
+ if username not in new_save_directory else \
+ new_save_directory.lstrip('/.')
+ print(f"Saved GGUF to https://huggingface.co/{link}")
+pass
+
+
+def patch_saving_functions(model):
+ import inspect
+ import re
+ import types
+ from typing import Callable, Optional, Union, List
+
+ # And now re add our saving methods!
+ if model.push_to_hub.__name__ == "unsloth_push_to_hub":
+ original_push_to_hub = model.original_push_to_hub
+ else:
+ original_push_to_hub = model.push_to_hub
+ pass
+
+ signature = str(inspect.signature(original_push_to_hub)).replace("NoneType", "None")
+ signature = signature[1:]
+ signature = re.sub("