diff --git a/.nojekyll b/.nojekyll index 89760cf9b..f3319dee8 100644 --- a/.nojekyll +++ b/.nojekyll @@ -1 +1 @@ -4eda9161 \ No newline at end of file +c81fd013 \ No newline at end of file diff --git a/docs/dataset-formats/index.html b/docs/dataset-formats/index.html index 2d8a081c2..6c2e1321c 100644 --- a/docs/dataset-formats/index.html +++ b/docs/dataset-formats/index.html @@ -363,7 +363,7 @@ Description - + Pre-training @@ -371,7 +371,7 @@ Description Data format for a pre-training completion task. - + Instruction Tuning @@ -379,7 +379,7 @@ Description Instruction tuning formats for supervised fine-tuning. - + Conversation @@ -387,7 +387,7 @@ Description Conversation format for supervised fine-tuning. - + Template-Free @@ -395,7 +395,7 @@ Description Construct prompts without a template. - + Custom Pre-Tokenized Dataset diff --git a/docs/input_output.html b/docs/input_output.html index 3993b314a..1362d9b1e 100644 --- a/docs/input_output.html +++ b/docs/input_output.html @@ -474,7 +474,7 @@ pre > code.sourceCode > span > a:first-child::before { text-decoration: underlin >>> print(tok.decode(row['input_ids'])) <s> Hello hi there!. goodbye farewell</s> -

We can check that the right tokens are ingored by comparing the labels to each token:

+

We can check that the right tokens are ignored by comparing the labels to each token:

import pandas as pd
 pd.DataFrame([{'token': tok.decode(i), 'label': l, 'id':i} for i,l in
               zip(row['input_ids'], row['labels'])])
diff --git a/docs/multimodal.html b/docs/multimodal.html new file mode 100644 index 000000000..c0ab7e631 --- /dev/null +++ b/docs/multimodal.html @@ -0,0 +1,766 @@ + + + + + + + + + +multimodal – Axolotl + + + + + + + + + + + + + + + + + + + + + + + + + + + +
+
+ + +
+ +
+ + +
+ + + +
+ + + + +
+

MultiModal / Vision Language Models (BETA)

+
+

Supported Models

+
    +
  • Mllama, i.e. llama with vision models
    • +
+
+
+

Usage

+

Currently multimodal support is limited and doesn’t have full feature parity. To finetune a multimodal Llama w/ LoRA, you’ll need to use the following in YAML in combination with the rest of the required hyperparams.

+
base_model: alpindale/Llama-3.2-11B-Vision-Instruct
+processor_type: AutoProcessor
+skip_prepare_dataset: true
+
+chat_template: llama3_2_vision
+datasets:
+  - path: HuggingFaceH4/llava-instruct-mix-vsft
+    type: chat_template
+    split: train[:1%]
+    field_messages: messages
+remove_unused_columns: false
+sample_packing: false
+
+# only finetune the Language model, leave the vision model and vision tower frozen
+lora_target_modules: 'language_model.model.layers.[\d]+.(mlp|cross_attn|self_attn).(up|down|gate|q|k|v|o)_proj'
+ + +
+
+ +
+ +
+ + + + + \ No newline at end of file diff --git a/search.json b/search.json index 98a59ecaa..81e8a718e 100644 --- a/search.json +++ b/search.json @@ -321,115 +321,87 @@ ] }, { - "objectID": "docs/faq.html", - "href": "docs/faq.html", - "title": "FAQ", + "objectID": "docs/multi-node.html", + "href": "docs/multi-node.html", + "title": "Multi Node", "section": "", - "text": "Q: The trainer stopped and hasn’t progressed in several minutes.\n\nA: Usually an issue with the GPUs communicating with each other. See the NCCL doc\n\nQ: Exitcode -9\n\nA: This usually happens when you run out of system RAM.\n\nQ: Exitcode -7 while using deepspeed\n\nA: Try upgrading deepspeed w: pip install -U deepspeed\n\nQ: AttributeError: ‘DummyOptim’ object has no attribute ‘step’\n\nA: You may be using deepspeed with single gpu. Please don’t set deepspeed: in yaml or cli.", - "crumbs": [ - "FAQ" - ] - }, - { - "objectID": "docs/config.html", - "href": "docs/config.html", - "title": "Config options", - "section": "", - "text": "# This is the huggingface model that contains *.pt, *.safetensors, or *.bin files\n# This can also be a relative path to a model on disk\nbase_model: ./llama-7b-hf\n# You can specify an ignore pattern if the model repo contains more than 1 model type (*.pt, etc)\nbase_model_ignore_patterns:\n# If the base_model repo on hf hub doesn't include configuration .json files,\n# You can set that here, or leave this empty to default to base_model\nbase_model_config: ./llama-7b-hf\n# You can specify to choose a specific model revision from huggingface hub\nrevision_of_model:\n# Optional tokenizer configuration path in case you want to use a different tokenizer\n# than the one defined in the base model\ntokenizer_config:\n# If you want to specify the type of model to load, AutoModelForCausalLM is a good choice too\nmodel_type: AutoModelForCausalLM\n# Corresponding tokenizer for the model AutoTokenizer is a good choice\ntokenizer_type: AutoTokenizer\n# Trust remote code for untrusted source\ntrust_remote_code:\n# use_fast option for tokenizer loading from_pretrained, default to True\ntokenizer_use_fast:\n# Whether to use the legacy tokenizer setting, defaults to True\ntokenizer_legacy:\n# Resize the model embeddings when new tokens are added to multiples of 32\n# This is reported to improve training speed on some models\nresize_token_embeddings_to_32x:\n\n# (Internal use only)\n# Used to identify which the model is based on\nis_falcon_derived_model:\nis_llama_derived_model:\nis_qwen_derived_model:\n# Please note that if you set this to true, `padding_side` will be set to \"left\" by default\nis_mistral_derived_model:\n\n# optional overrides to the base model configuration\noverrides_of_model_config:\n # RoPE Scaling https://github.com/huggingface/transformers/pull/24653\n rope_scaling:\n type: # linear | dynamic\n factor: # float\n\n# optional overrides to the bnb 4bit quantization configuration\n# https://huggingface.co/docs/transformers/main/main_classes/quantization#transformers.BitsAndBytesConfig\nbnb_config_kwargs:\n # These are default values\n llm_int8_has_fp16_weight: false\n bnb_4bit_quant_type: nf4\n bnb_4bit_use_double_quant: true\n\n\n# Whether you are training a 4-bit GPTQ quantized model\ngptq: true\n\n# This will attempt to quantize the model down to 8 bits and use adam 8 bit optimizer\nload_in_8bit: true\n# Use bitsandbytes 4 bit\nload_in_4bit:\n\n# Use CUDA bf16\nbf16: true # bool or 'full' for `bf16_full_eval`. require >=ampere\n# Use CUDA fp16\nfp16: true\n# Use CUDA tf32\ntf32: true # require >=ampere\n\n# No AMP (automatic mixed precision)\nbfloat16: true # require >=ampere\nfloat16: true\n\n# Limit the memory for all available GPUs to this amount (if an integer, expressed in gigabytes); default: unset\ngpu_memory_limit: 20GiB\n# Do the LoRA/PEFT loading on CPU -- this is required if the base model is so large it takes up most or all of the available GPU VRAM, e.g. during a model and LoRA merge\nlora_on_cpu: true\n\n# A list of one or more datasets to finetune the model with\ndatasets:\n # HuggingFace dataset repo | s3://,gs:// path | \"json\" for local dataset, make sure to fill data_files\n - path: vicgalle/alpaca-gpt4\n # The type of prompt to use for training. [alpaca, sharegpt, gpteacher, oasst, reflection]\n type: alpaca # format | format:<prompt_style> (chat/instruct) | <prompt_strategies>.load_<load_fn>\n ds_type: # Optional[str] (json|arrow|parquet|text|csv) defines the datatype when path is a file\n data_files: # Optional[str] path to source data files\n shards: # Optional[int] number of shards to split data into\n name: # Optional[str] name of dataset configuration to load\n train_on_split: train # Optional[str] name of dataset split to load from\n\n # Optional[str] fastchat conversation type, only used with type: sharegpt\n conversation: # Options (see Conversation 'name'): https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py\n field_human: # Optional[str]. Human key to use for conversation.\n field_model: # Optional[str]. Assistant key to use for conversation.\n # Add additional keys from your dataset as input or output roles\n roles:\n input: # Optional[List[str]]. These will be masked based on train_on_input\n output: # Optional[List[str]].\n\n # Custom user instruction prompt\n - path: repo\n type:\n # The below are defaults. only set what's needed if you use a different column name.\n system_prompt: \"\"\n system_format: \"{system}\"\n field_system: system\n field_instruction: instruction\n field_input: input\n field_output: output\n\n # Customizable to be single line or multi-line\n # Use {instruction}/{input} as key to be replaced\n # 'format' can include {input}\n format: |-\n User: {instruction} {input}\n Assistant:\n # 'no_input_format' cannot include {input}\n no_input_format: \"{instruction} \"\n\n # For `completion` datsets only, uses the provided field instead of `text` column\n field:\n\n# If false, the datasets will not be shuffled and will keep their original order in `datasets`.\n# The same applies to the `test_datasets` option and the `pretraining_dataset` option. Default is true.\nshuffle_merged_datasets: true\n\n# A list of one or more datasets to eval the model with.\n# You can use either test_datasets, or val_set_size, but not both.\ntest_datasets:\n - path: /workspace/data/eval.jsonl\n ds_type: json\n # You need to specify a split. For \"json\" datasets the default split is called \"train\".\n split: train\n type: completion\n data_files:\n - /workspace/data/eval.jsonl\n\n# use RL training: 'dpo', 'ipo', 'kto'\nrl:\n\n# Saves the desired chat template to the tokenizer_config.json for easier inferencing\n# Currently supports chatml and inst (mistral/mixtral)\nchat_template: chatml\n# Changes the default system message\ndefault_system_message: You are a helpful assistant. Please give a long and detailed answer. # Currently only supports chatml.\n# Axolotl attempts to save the dataset as an arrow after packing the data together so\n# subsequent training attempts load faster, relative path\ndataset_prepared_path: data/last_run_prepared\n# Push prepared dataset to hub\npush_dataset_to_hub: # repo path\n# The maximum number of processes to use while preprocessing your input dataset. This defaults to `os.cpu_count()`\n# if not set.\ndataset_processes: # defaults to os.cpu_count() if not set\n# Keep dataset in memory while preprocessing\n# Only needed if cached dataset is taking too much storage\ndataset_keep_in_memory:\n# push checkpoints to hub\nhub_model_id: # private repo path to push finetuned model\n# how to push checkpoints to hub\n# https://huggingface.co/docs/transformers/v4.31.0/en/main_classes/trainer#transformers.TrainingArguments.hub_strategy\nhub_strategy:\n# Whether to use hf `use_auth_token` for loading datasets. Useful for fetching private datasets\n# Required to be true when used in combination with `push_dataset_to_hub`\nhf_use_auth_token: # boolean\n# How much of the dataset to set aside as evaluation. 1 = 100%, 0.50 = 50%, etc. 0 for no eval.\nval_set_size: 0.04\n# Num shards for whole dataset\ndataset_shard_num:\n# Index of shard to use for whole dataset\ndataset_shard_idx:\n\n# The maximum length of an input to train with, this should typically be less than 2048\n# as most models have a token/context limit of 2048\nsequence_len: 2048\n# Pad inputs so each step uses constant sized buffers\n# This will reduce memory fragmentation and may prevent OOMs, by re-using memory more efficiently\npad_to_sequence_len:\n# Use efficient multi-packing with block diagonal attention and per sequence position_ids. Recommend set to 'true'\nsample_packing:\n# Set to 'false' if getting errors during eval with sample_packing on.\neval_sample_packing:\n# You can set these packing optimizations AFTER starting a training at least once.\n# The trainer will provide recommended values for these values.\nsample_packing_eff_est:\ntotal_num_tokens:\n# Increasing the following values helps with packing, but usually only slightly (<%1.)\n# The number of samples packed at a time.\nsample_packing_group_size: 100000\n# The number of samples which can be packed into one sequence. Increase if using a large sequence_len with many short samples.\nsample_packing_bin_size: 200\n\n# Passed through to transformers when loading the model when launched without accelerate\n# Use `sequential` when training w/ model parallelism to limit memory\ndevice_map:\n# Defines the max memory usage per gpu on the system. Passed through to transformers when loading the model.\nmax_memory:\n\n# If you want to use 'lora' or 'qlora' or leave blank to train all parameters in original model\nadapter: lora\n# If you already have a lora model trained that you want to load, put that here.\n# This means after training, if you want to test the model, you should set this to the value of `output_dir`.\n# Note that if you merge an adapter to the base model, a new subdirectory `merged` will be created under the `output_dir`.\nlora_model_dir:\n\n# LoRA hyperparameters\n# For more details about the following options, see:\n# https://www.anyscale.com/blog/fine-tuning-llms-lora-or-full-parameter-an-in-depth-analysis-with-llama-2\nlora_r: 8\nlora_alpha: 16\nlora_dropout: 0.05\nlora_target_modules:\n - q_proj\n - v_proj\n# - k_proj\n# - o_proj\n# - gate_proj\n# - down_proj\n# - up_proj\nlora_target_linear: # If true, will target all linear modules\npeft_layers_to_transform: # The layer indices to transform, otherwise, apply to all layers\n\n# If you added new tokens to the tokenizer, you may need to save some LoRA modules because they need to know the new tokens.\n# For LLaMA and Mistral, you need to save `embed_tokens` and `lm_head`. It may vary for other models.\n# `embed_tokens` converts tokens to embeddings, and `lm_head` converts embeddings to token probabilities.\n# https://github.com/huggingface/peft/issues/334#issuecomment-1561727994\nlora_modules_to_save:\n# - embed_tokens\n# - lm_head\n\nlora_fan_in_fan_out: false\n\n# LoRA+ hyperparameters\n# For more details about the following options, see:\n# https://arxiv.org/abs/2402.12354 and `src/axolotl/core/train_builder.py`\nloraplus_lr_ratio: # loraplus learning rate ratio lr_B / lr_A. Recommended value is 2^4.\nloraplus_lr_embedding: # loraplus learning rate for lora embedding layers. Default value is 1e-6.\n\npeft:\n # Configuration options for loftq initialization for LoRA\n # https://huggingface.co/docs/peft/developer_guides/quantization#loftq-initialization\n loftq_config:\n loftq_bits: # typically 4 bits\n\n# ReLoRA configuration\n# Must use either 'lora' or 'qlora' adapter, and does not support fsdp or deepspeed\nrelora_steps: # Number of steps per ReLoRA restart\nrelora_warmup_steps: # Number of per-restart warmup steps\nrelora_anneal_steps: # Number of anneal steps for each relora cycle\nrelora_prune_ratio: # threshold for optimizer magnitude when pruning\nrelora_cpu_offload: # True to perform lora weight merges on cpu during restarts, for modest gpu memory savings\n\n# wandb configuration if you're using it\n# Make sure your `WANDB_API_KEY` environment variable is set (recommended) or you login to wandb with `wandb login`.\nwandb_mode: # \"offline\" to save run metadata locally and not sync to the server, \"disabled\" to turn off wandb\nwandb_project: # Your wandb project name\nwandb_entity: # A wandb Team name if using a Team\nwandb_watch:\nwandb_name: # Set the name of your wandb run\nwandb_run_id: # Set the ID of your wandb run\nwandb_log_model: # \"checkpoint\" to log model to wandb Artifacts every `save_steps` or \"end\" to log only at the end of training\n\n# mlflow configuration if you're using it\nmlflow_tracking_uri: # URI to mlflow\nmlflow_experiment_name: # Your experiment name\nhf_mlflow_log_artifacts: # set to true to copy each saved checkpoint on each save to mlflow artifact registry\n\n# Where to save the full-finetuned model to\noutput_dir: ./completed-model\n\n# Whether to use torch.compile and which backend to use\ntorch_compile: # bool\ntorch_compile_backend: # Optional[str]\n\n# Training hyperparameters\n\n# If greater than 1, backpropagation will be skipped and the gradients will be accumulated for the given number of steps.\ngradient_accumulation_steps: 1\n# The number of samples to include in each batch. This is the number of samples sent to each GPU.\n# Batch size per gpu = micro_batch_size * gradient_accumulation_steps\nmicro_batch_size: 2\neval_batch_size:\nnum_epochs: 4\nwarmup_steps: 100 # cannot use with warmup_ratio\nwarmup_ratio: 0.05 # cannot use with warmup_steps\nlearning_rate: 0.00003\nlr_quadratic_warmup:\nlogging_steps:\neval_steps: # Leave empty to eval at each epoch, integers for every N steps. decimal for fraction of total steps\nevals_per_epoch: # number of times per epoch to run evals, mutually exclusive with eval_steps\nsave_strategy: # Set to `\"no\"` to skip checkpoint saves\nsave_steps: # Leave empty to save at each epoch\nsaves_per_epoch: # number of times per epoch to save a checkpoint, mutually exclusive with save_steps\nsave_total_limit: # Checkpoints saved at a time\n# Maximum number of iterations to train for. It precedes num_epochs which means that\n# if both are set, num_epochs will not be guaranteed.\n# e.g., when 1 epoch is 1000 steps => `num_epochs: 2` and `max_steps: 100` will train for 100 steps\nmax_steps:\n\neval_table_size: # Approximate number of predictions sent to wandb depending on batch size. Enabled above 0. Default is 0\neval_max_new_tokens: # Total number of tokens generated for predictions sent to wandb. Default is 128\neval_causal_lm_metrics: # HF evaluate metrics used during evaluation. Default is [\"sacrebleu\", \"comet\", \"ter\", chrf]\n\nloss_watchdog_threshold: # High loss value, indicating the learning has broken down (a good estimate is ~2 times the loss at the start of training)\nloss_watchdog_patience: # Number of high-loss steps in a row before the trainer aborts (default: 3)\n\n# Save model as safetensors (require safetensors package)\nsave_safetensors:\n\n# Whether to mask out or include the human's prompt from the training labels\ntrain_on_inputs: false\n# Group similarly sized data to minimize padding.\n# May be slower to start, as it must download and sort the entire dataset.\n# Note that training loss may have an oscillating pattern with this enabled.\ngroup_by_length: false\n\n# Whether to use gradient checkpointing https://huggingface.co/docs/transformers/v4.18.0/en/performance#gradient-checkpointing\ngradient_checkpointing: false\n# additional kwargs to pass to the trainer for gradient checkpointing\n# gradient_checkpointing_kwargs:\n# use_reentrant: true\n\n# Stop training after this many evaluation losses have increased in a row\n# https://huggingface.co/transformers/v4.2.2/_modules/transformers/trainer_callback.html#EarlyStoppingCallback\nearly_stopping_patience: 3\n\n# Specify a scheduler and kwargs to use with the optimizer\nlr_scheduler: # 'one_cycle' | 'log_sweep' | empty for cosine\nlr_scheduler_kwargs:\ncosine_min_lr_ratio: # decay lr to some percentage of the peak lr, e.g. cosine_min_lr_ratio=0.1 for 10% of peak lr\ncosine_constant_lr_ratio: # freeze lr at some percentage of the step, e.g. cosine_constant_lr_ratio=0.8 means start cosine_min_lr at 80% of training step (https://arxiv.org/pdf/2308.04014.pdf)\n\n# For one_cycle optim\nlr_div_factor: # Learning rate div factor\n\n# Specify optimizer\n# Valid values are driven by the Transformers OptimizerNames class, see:\n# https://github.com/huggingface/transformers/blob/95b374952dc27d8511541d6f5a4e22c9ec11fb24/src/transformers/training_args.py#L134\n#\n# Note that not all optimizers may be available in your environment, ex: 'adamw_anyprecision' is part of\n# torchdistx, 'adamw_bnb_8bit' is part of bnb.optim.Adam8bit, etc. When in doubt, it is recommended to start with the optimizer used\n# in the examples/ for your model and fine-tuning use case.\n#\n# Valid values for 'optimizer' include:\n# - adamw_hf\n# - adamw_torch\n# - adamw_torch_fused\n# - adamw_torch_xla\n# - adamw_apex_fused\n# - adafactor\n# - adamw_anyprecision\n# - sgd\n# - adagrad\n# - adamw_bnb_8bit\n# - lion_8bit\n# - lion_32bit\n# - paged_adamw_32bit\n# - paged_adamw_8bit\n# - paged_lion_32bit\n# - paged_lion_8bit\n# - galore_adamw\n# - galore_adamw_8bit\n# - galore_adafactor\n# - galore_adamw_layerwise\n# - galore_adamw_8bit_layerwise\n# - galore_adafactor_layerwise\noptimizer:\n# Dictionary of arguments to pass to the optimizer\noptim_args:\n# For Galore Optimizers the following optim_args are available\n# rank: # type: int\n# update_proj_gap # type: int\n# scale # type: float\n# proj_type: # type: str, default = std\n\n# The target modules to optimize, i.e. the module names that you would like to train, right now this is used only for GaLore algorithm\noptim_target_modules:\n# - self_attn # for llama\n# - mlp\n\n# Specify weight decay\nweight_decay:\n# adamw hyperparams\nadam_beta1:\nadam_beta2:\nadam_epsilon:\n# Gradient clipping max norm\nmax_grad_norm:\n\n# Augmentation techniques\n# NEFT https://arxiv.org/abs/2310.05914, set this to a number (paper default is 5) to add noise to embeddings\n# currently only supported on Llama and Mistral\nneftune_noise_alpha:\n\n# Whether to bettertransformers\nflash_optimum:\n# Whether to use xformers attention patch https://github.com/facebookresearch/xformers:\nxformers_attention:\n# Whether to use flash attention patch https://github.com/Dao-AILab/flash-attention:\nflash_attention:\nflash_attn_cross_entropy: # Whether to use flash-attention cross entropy implementation - advanced use only\nflash_attn_rms_norm: # Whether to use flash-attention rms norm implementation - advanced use only\nflash_attn_fuse_qkv: # Whether to fuse QKV into a single operation\nflash_attn_fuse_mlp: # Whether to fuse part of the MLP into a single operation\n# Whether to use scaled-dot-product attention\n# https://pytorch.org/docs/stable/generated/torch.nn.functional.scaled_dot_product_attention.html\nsdp_attention:\n# Shifted-sparse attention (only llama) - https://arxiv.org/pdf/2309.12307.pdf\ns2_attention:\n# Resume from a specific checkpoint dir\nresume_from_checkpoint:\n# If resume_from_checkpoint isn't set and you simply want it to start where it left off.\n# Be careful with this being turned on between different models.\nauto_resume_from_checkpoints: false\n\n# Don't mess with this, it's here for accelerate and torchrun\nlocal_rank:\n\n# Add or change special tokens.\n# If you add tokens here, you don't need to add them to the `tokens` list.\nspecial_tokens:\n # bos_token: \"<s>\"\n # eos_token: \"</s>\"\n # unk_token: \"<unk>\"\n # pad_token: \"[PAD]\"\n\n# Add extra tokens.\ntokens:\n\n# FSDP\nfsdp:\nfsdp_config:\n\n# Deepspeed config path. e.g., deepspeed_configs/zero3.json\ndeepspeed:\n\n# Advanced DDP Arguments\nddp_timeout:\nddp_bucket_cap_mb:\nddp_broadcast_buffers:\n\n# Path to torch distx for optim 'adamw_anyprecision'\ntorchdistx_path:\n\n# Set to HF dataset for type: 'completion' for streaming instead of pre-tokenize\npretraining_dataset:\n\n# Debug mode\ndebug:\n\n# Seed\nseed:\n\n# Allow overwrite yml config using from cli\nstrict:", - "crumbs": [ - "Reference", - "Config options" - ] - }, - { - "objectID": "docs/amd_hpc.html", - "href": "docs/amd_hpc.html", - "title": "Training with AMD GPUs on HPC Systems", - "section": "", - "text": "This guide provides step-by-step instructions for installing and configuring Axolotl on a High-Performance Computing (HPC) environment equipped with AMD GPUs.", + "text": "You will need to create a configuration for accelerate, either by using accelerate config and follow the instructions or you can use one of the preset below:\n~/.cache/huggingface/accelerate/default_config.yaml\nConfigure your model to use FSDP with for example:", "crumbs": [ "How-To Guides", - "Training with AMD GPUs on HPC Systems" + "Multi Node" ] }, { - "objectID": "docs/amd_hpc.html#setup", - "href": "docs/amd_hpc.html#setup", - "title": "Training with AMD GPUs on HPC Systems", - "section": "Setup", - "text": "Setup\n\n1. Install Python\nWe recommend using Miniforge, a minimal conda-based Python distribution:\ncurl -L -O \"https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh\"\nbash Miniforge3-$(uname)-$(uname -m).sh\n\n\n2. Configure Python Environment\nAdd Python to your PATH and ensure it’s available at login:\necho 'export PATH=~/miniforge3/bin:$PATH' >> ~/.bashrc\necho 'if [ -f ~/.bashrc ]; then . ~/.bashrc; fi' >> ~/.bash_profile\n\n\n3. Load AMD GPU Software\nLoad the ROCm module:\nmodule load rocm/5.7.1\nNote: The specific module name and version may vary depending on your HPC system. Consult your system documentation for the correct module name.\n\n\n4. Install PyTorch\nInstall PyTorch with ROCm support:\npip install -U torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.7 --force-reinstall\n\n\n5. Install Flash Attention\nClone and install the Flash Attention repository:\ngit clone --recursive https://github.com/ROCmSoftwarePlatform/flash-attention.git\nexport GPU_ARCHS=\"gfx90a\"\ncd flash-attention\nexport PYTHON_SITE_PACKAGES=$(python -c 'import site; print(site.getsitepackages()[0])')\npatch \"${PYTHON_SITE_PACKAGES}/torch/utils/hipify/hipify_python.py\" hipify_patch.patch\npip install .\n\n\n6. Install Axolotl\nClone and install Axolotl:\ngit clone https://github.com/axolotl-ai-cloud/axolotl\ncd axolotl\npip install packaging ninja\npip install -e .\n\n\n7. Apply xformers Workaround\nxformers appears to be incompatible with ROCm. Apply the following workarounds: - Edit $HOME/packages/axolotl/src/axolotl/monkeypatch/llama_attn_hijack_flash.py modifying the code to always return False for SwiGLU availability from xformers. - Edit $HOME/miniforge3/lib/python3.10/site-packages/xformers/ops/swiglu_op.py replacing the “SwiGLU” function with a pass statement.\n\n\n8. Prepare Job Submission Script\nCreate a script for job submission using your HPC’s particular software (e.g. Slurm, PBS). Include necessary environment setup and the command to run Axolotl training. If the compute node(s) do(es) not have internet access, it is recommended to include\nexport TRANSFORMERS_OFFLINE=1\nexport HF_DATASETS_OFFLINE=1\n\n\n9. Download Base Model\nDownload a base model using the Hugging Face CLI:\nhuggingface-cli download meta-llama/Meta-Llama-3.1-8B --local-dir ~/hfdata/llama3.1-8B\n\n\n10. Create Axolotl Configuration\nCreate an Axolotl configuration file (YAML format) tailored to your specific training requirements and dataset. Use FSDP for multi-node training.\nNote: Deepspeed did not work at the time of testing. However, if anyone managed to get it working, please let us know.\n\n\n11. Preprocess Data\nRun preprocessing on the login node:\nCUDA_VISIBLE_DEVICES=\"\" python -m axolotl.cli.preprocess /path/to/your/config.yaml\n\n\n12. Train\nYou are now ready to submit your previously prepared job script. 🚂", + "objectID": "docs/multi-node.html#machine-configuration", + "href": "docs/multi-node.html#machine-configuration", + "title": "Multi Node", + "section": "Machine configuration", + "text": "Machine configuration\nOn each machine you need a copy of Axolotl, we suggest using the same commit to ensure compatibility.\nYou will also need to have the same configuration file for your model on each machine.\nOn the main machine only, make sure the port you set as main_process_port is open in TCP and reachable by other machines.\nAll you have to do now is launch using accelerate as you would usually do on each machine and voila, the processes will start once you have launched accelerate on every machine.", "crumbs": [ "How-To Guides", - "Training with AMD GPUs on HPC Systems" + "Multi Node" ] }, { - "objectID": "docs/input_output.html", - "href": "docs/input_output.html", - "title": "Template-free prompt construction", + "objectID": "docs/mac.html", + "href": "docs/mac.html", + "title": "Mac M-series", "section": "", - "text": "Background\n\nMasking Inputs\nYou may not want prompt templates\nThe input_output format\n\nUsage\n\n1. Prepare Data\n2. Use type: input_output\n3. Check the prompts", + "text": "Currently Axolotl on Mac is partially usable, many of the dependencies of Axolotl including Pytorch do not support MPS or have incomplete support.\nCurrent support:\n\nSupport for all models\nFull training of models\nLoRA training\nSample packing\nFP16 and BF16 (awaiting AMP support for MPS in Pytorch)\nTri-dao’s flash-attn (until it is supported use spd_attention as an alternative)\nxformers\nbitsandbytes (meaning no 4/8 bits loading and bnb optimizers)\nqlora\nDeepSpeed\n\nUntested: - FSDP", "crumbs": [ "How-To Guides", - "Template-free prompt construction" + "Mac M-series" ] }, { - "objectID": "docs/input_output.html#background", - "href": "docs/input_output.html#background", - "title": "Template-free prompt construction", - "section": "Background", - "text": "Background\n\n\nMasking Inputs\nOne of the most popular features of axolotl is setting the following configuration value:\ntrain_on_inputs: false\nIf you declare a dataset formats such as alpaca or chatml, axolotl knows what is an input (i.e. human) vs. an output (i.e. the assistant) and masks the input labels so that your model can focus on predicting the outputs only.\n\n\n\nYou may not want prompt templates\nHowever, there are many situations where you don’t want to use one of these formats or templates. This is because they can:\n\nAdd unnecessary boilerplate to your prompts.\nCreate artifacts like special delimiters <|im_start|> that can quickly become footguns if you don’t include them correctly at inference time.\nEnforce a chat interface when you do not want one. Sometimes you just want to fine-tune a model to a very specific task and do NOT want multi-turn conversations, roles, etc.\nLimit you to only certain roles that the template allows.\n\n\n\n\nThe input_output format\nYou can construct your prompts without a template by using the input_output format, by setting type: input_output in your configuration file like this:\nconfig.yml\ntrain_on_inputs: false # Mask segments of your data\ndatasets:\n - path: output.jsonl\n type: input_output # use template free prompt construction\nUnlike type: completion, which is also template-free, type: input_output allows you to mask segments of your text. More details on how this works are described below.", - "crumbs": [ - "How-To Guides", - "Template-free prompt construction" - ] - }, - { - "objectID": "docs/input_output.html#usage", - "href": "docs/input_output.html#usage", - "title": "Template-free prompt construction", - "section": "Usage", - "text": "Usage\nThis is how you can use the input_output format:\n\n\n1. Prepare Data\nTo use the input_output format, collect your data in the following format into a jsonl file (below is the first row from the file output.jsonl` pretty printed):\n$ head -n1 output.jsonl | python -m json.tool\n\n{\n \"segments\": [\n {\n \"label\": true,\n \"text\": \"<s>Hello\\n\"\n },\n {\n \"label\": true,\n \"text\": \"hi there!. \"\n },\n {\n \"label\": false,\n \"text\": \"goodbye \"\n },\n {\n \"label\": true,\n \"text\": \"farewell</s>\"\n }\n ]\n}\n\nSet label:false when you want to mask a segment of text so that the model isn’t trained on it. Some things to keep in mind:\n\n[!IMPORTANT] 1. EOS, BOS, spaces, newlines etc. are entirely up to you. Axolotl concatenates all the segments as-is. The tokenizer doesn’t add anything additional. Notice how I added spaces, newlines, <s> (BOS), and </s> (EOS) myself. 2. Make sure you check the materialized output to validate that the prompt is getting assembled how you like.\n\n\n\n\n2. Use type: input_output\nLet’s materialize data with our output.jsonl file by setting type: input_output in our axolotl config:\n# training_config.yaml\nbase_model: mistralai/Mistral-7B-v0.1\ndata_seed: 49\nseed: 49\n\ndatasets:\n - path: output.jsonl\n type: input_output\nval_set_size: 0.1\n\nsequence_len: 896\nsample_packing: false\n\nmicro_batch_size: 2\ngradient_accumulation_steps: 3\neval_batch_size: 2\nnum_epochs: 1\nlearning_rate: 0.0002\n\ntrain_on_inputs: false\nspecial_tokens:\n bos_token: \"<s>\"\n eos_token: \"</s>\"\n unk_token: \"<unk>\"\nYou can use the following command to materialize your data. The --debug flag will print the tokens, along with the labels so you can verify that the correct items are being ignored:\n$ python -m axolotl.cli.preprocess training_config.yaml --debug\n\n...\n[2024-03-05 23:36:46,969] [INFO] [axolotl.check_example_labels:35] [PID:607731] [RANK:0] <s>(1, 1) Hello(22557, 22557)\n(13, 13) hi(12014, 12014) there(736, 736) !(28808, 28808) .(28723, 28723) (28705, 28705) good(-100, 1179) bye(-100, 17664) (-100, 28705) fare(19111, 19111) well(5458, 5458) </s>(2, 2)\nThe format is decoded_token(label, token_id), for example, <s>(1, 1) means that the token is <s>, the label is 1 and the token_id is 1. When the label is -100 then that token is ignored for training.\n\n\n\n3. Check the prompts\nHere is another way to check the materialized output:\nfrom transformers import AutoTokenizer\nfrom datasets import load_from_disk\nimport yaml\n\ndirectory = !ls last_run_prepared/\nwith open('training_config.yaml', 'r') as f:\n cfg = yaml.safe_load(f)\nmodel_id = cfg['base_model']\ntok = AutoTokenizer.from_pretrained(model_id)\nds = load_from_disk(f'last_run_prepared/{directory[0]}/')\n>>> row = ds[0]\n>>> print(tok.decode(row['input_ids']))\n<s> Hello\n hi there!. goodbye farewell</s>\nWe can check that the right tokens are ingored by comparing the labels to each token:\nimport pandas as pd\npd.DataFrame([{'token': tok.decode(i), 'label': l, 'id':i} for i,l in\n zip(row['input_ids'], row['labels'])])\n\n\n\ntoken\nlabel\nid\n\n\n\n\n0\n<s>\n1\n\n\n1\nHello\n22557\n\n\n2\n\\n\n13\n\n\n3\nhi\n12014\n\n\n4\nthere\n736\n\n\n5\n!\n28808\n\n\n6\n.\n28723\n\n\n7\n\n28705\n\n\n8\ngood\n-100\n\n\n9\nbye\n-100\n\n\n10\n\n-100\n\n\n11\nfare\n19111\n\n\n12\nwell\n5458\n\n\n13\n</s>\n2\n\n\n\nIf we look at the input data, the above table seems correct! (The jsonl version is repeated below for reference):\n$ head -n1 output.jsonl | python -m json.tool\n\n{\n \"segments\": [\n {\n \"label\": true,\n \"text\": \"<s>Hello\\n\"\n },\n {\n \"label\": true,\n \"text\": \"hi there!. \"\n },\n {\n \"label\": false,\n \"text\": \"goodbye \"\n },\n {\n \"label\": true,\n \"text\": \"farewell</s>\"\n }\n ]\n}", - "crumbs": [ - "How-To Guides", - "Template-free prompt construction" - ] - }, - { - "objectID": "examples/colab-notebooks/colab-axolotl-example.html", - "href": "examples/colab-notebooks/colab-axolotl-example.html", - "title": "Example notebook for running Axolotl on google colab", + "objectID": "docs/rlhf.html", + "href": "docs/rlhf.html", + "title": "RLHF (Beta)", "section": "", - "text": "import torch\n# Check so there is a gpu available, a T4(free tier) is enough to run this notebook\nassert (torch.cuda.is_available()==True)" + "text": "Overview\nReinforcement 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:\n\nProximal Policy Optimization (PPO) (not yet supported in axolotl)\nDirect Preference Optimization (DPO)\nIdentity Preference Optimization (IPO)\n\n\n\nRLHF using Axolotl\n\n[!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.\n\nThe 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\n\nDPO\nrl: dpo\ndatasets:\n - path: Intel/orca_dpo_pairs\n split: train\n type: chatml.intel\n - path: argilla/ultrafeedback-binarized-preferences\n split: train\n type: chatml.argilla\n\n\nIPO\nrl: ipo\n\n\nORPO\nPaper: https://arxiv.org/abs/2403.07691\nrl: orpo\norpo_alpha: 0.1\nremove_unused_columns: false\n\nchat_template: chatml\ndatasets:\n - path: argilla/ultrafeedback-binarized-preferences-cleaned\n type: chat_template.argilla\n\n\nUsing local dataset files\ndatasets:\n - ds_type: json\n data_files:\n - orca_rlhf.jsonl\n split: train\n type: chatml.intel\n\n\nTrl autounwrap for peft\nTrl 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.\n# load ref model when adapter training.\nrl_adapter_ref_model: true", + "crumbs": [ + "How-To Guides", + "RLHF (Beta)" + ] }, { - "objectID": "examples/colab-notebooks/colab-axolotl-example.html#install-axolotl-and-dependencies", - "href": "examples/colab-notebooks/colab-axolotl-example.html#install-axolotl-and-dependencies", - "title": "Example notebook for running Axolotl on google colab", - "section": "Install Axolotl and dependencies", - "text": "Install Axolotl and dependencies\n\n!pip install -e git+https://github.com/axolotl-ai-cloud/axolotl#egg=axolotl\n!pip install flash-attn==\"2.5.0\"\n!pip install deepspeed==\"0.13.1\"!pip install mlflow==\"2.13.0\"" + "objectID": "docs/unsloth.html", + "href": "docs/unsloth.html", + "title": "Unsloth", + "section": "", + "text": "Overview\nUnsloth provides hand-written optimized kernels for LLM finetuning that slightly improve speed and VRAM over standard industry baselines.\n\n\nInstallation\nThe following will install unsloth from source and downgrade xformers as unsloth is incompatible with the most up to date libraries.\npip install --no-deps \"unsloth @ git+https://github.com/unslothai/unsloth.git\"\npip install --no-deps --force-reinstall xformers==0.0.26.post1\n\n\nUsing unsloth w Axolotl\nAxolotl exposes a few configuration options to try out unsloth and get most of the performance gains.\nOur unsloth integration is currently limited to the following model architectures: - llama\nThese options are specific to LoRA finetuning and cannot be used for multi-GPU finetuning\nunsloth_lora_mlp: true\nunsloth_lora_qkv: true\nunsloth_lora_o: true\nThese options are composable and can be used with multi-gpu finetuning\nunsloth_cross_entropy_loss: true\nunsloth_rms_norm: true\nunsloth_rope: true\n\n\nLimitations\n\nSingle GPU only; e.g. no multi-gpu support\nNo deepspeed or FSDP support (requires multi-gpu)\nLoRA + QLoRA support only. No full fine tunes or fp8 support.\nLimited model architecture support. Llama, Phi, Gemma, Mistral only\nNo MoE support.", + "crumbs": [ + "How-To Guides", + "Unsloth" + ] }, { - "objectID": "examples/colab-notebooks/colab-axolotl-example.html#create-an-yaml-config-file", - "href": "examples/colab-notebooks/colab-axolotl-example.html#create-an-yaml-config-file", - "title": "Example notebook for running Axolotl on google colab", - "section": "Create an yaml config file", - "text": "Create an yaml config file\n\nimport yaml\n\n# Your YAML string\nyaml_string = \"\"\"\nbase_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T\nmodel_type: LlamaForCausalLM\ntokenizer_type: LlamaTokenizer\n\nload_in_8bit: false\nload_in_4bit: true\nstrict: false\n\ndatasets:\n - path: mhenrichsen/alpaca_2k_test\n type: alpaca\ndataset_prepared_path:\nval_set_size: 0.05\noutput_dir: ./outputs/qlora-out\n\nadapter: qlora\nlora_model_dir:\n\nsequence_len: 4096\nsample_packing: true\neval_sample_packing: false\npad_to_sequence_len: true\n\nlora_r: 32\nlora_alpha: 16\nlora_dropout: 0.05\nlora_target_modules:\nlora_target_linear: true\nlora_fan_in_fan_out:\n\nwandb_project:\nwandb_entity:\nwandb_watch:\nwandb_name:\nwandb_log_model:\n\ngradient_accumulation_steps: 4\nmicro_batch_size: 2\nnum_epochs: 4\noptimizer: paged_adamw_32bit\nlr_scheduler: cosine\nlearning_rate: 0.0002\n\ntrain_on_inputs: false\ngroup_by_length: false\nbf16: auto\nfp16:\ntf32: false\n\ngradient_checkpointing: true\nearly_stopping_patience:\nresume_from_checkpoint:\nlocal_rank:\nlogging_steps: 1\nxformers_attention:\nflash_attention: true\n\nwarmup_steps: 10\nevals_per_epoch: 4\nsaves_per_epoch: 1\ndebug:\ndeepspeed:\nweight_decay: 0.0\nfsdp:\nfsdp_config:\nspecial_tokens:\n\n\"\"\"\n\n# Convert the YAML string to a Python dictionary\nyaml_dict = yaml.safe_load(yaml_string)\n\n# Specify your file path\nfile_path = 'test_axolotl.yaml'\n\n# Write the YAML file\nwith open(file_path, 'w') as file:\n yaml.dump(yaml_dict, file)" + "objectID": "TODO.html", + "href": "TODO.html", + "title": "todo list", + "section": "", + "text": "[] Validation of parameters for combinations that won’t work\n\n\n\n\nFSDP offload and gradient_checkpointing - https://github.com/pytorch/pytorch/issues/82203\nadamw_bnb_8bit doesn’t play well with FSDP offload" }, { - "objectID": "examples/colab-notebooks/colab-axolotl-example.html#launch-the-training", - "href": "examples/colab-notebooks/colab-axolotl-example.html#launch-the-training", - "title": "Example notebook for running Axolotl on google colab", - "section": "Launch the training", - "text": "Launch the training\n\n# By using the ! the comand will be executed as a bash command\n!accelerate launch -m axolotl.cli.train /content/test_axolotl.yaml" + "objectID": "TODO.html#things-that-are-known-not-to-work", + "href": "TODO.html#things-that-are-known-not-to-work", + "title": "todo list", + "section": "", + "text": "FSDP offload and gradient_checkpointing - https://github.com/pytorch/pytorch/issues/82203\nadamw_bnb_8bit doesn’t play well with FSDP offload" }, { - "objectID": "examples/colab-notebooks/colab-axolotl-example.html#play-with-inference", - "href": "examples/colab-notebooks/colab-axolotl-example.html#play-with-inference", - "title": "Example notebook for running Axolotl on google colab", - "section": "Play with inference", - "text": "Play with inference\n\n# By using the ! the comand will be executed as a bash command\n!accelerate launch -m axolotl.cli.inference /content/test_axolotl.yaml \\\n --qlora_model_dir=\"./qlora-out\" --gradio" + "objectID": "FAQS.html", + "href": "FAQS.html", + "title": "FAQs", + "section": "", + "text": "FAQs\n\nCan you train StableLM with this? Yes, but only with a single GPU atm. Multi GPU support is coming soon! Just waiting on this PR\nWill this work with Deepspeed? That’s still a WIP, but setting export ACCELERATE_USE_DEEPSPEED=true should work in some cases\nError invalid argument at line 359 in file /workspace/bitsandbytes/csrc/pythonInterface.c /arrow/cpp/src/arrow/filesystem/s3fs.cc:2598: arrow::fs::FinalizeS3 was not called even though S3 was initialized. This could lead to a segmentation fault at exit. Try reinstalling bitsandbytes and transformers from source." + }, + { + "objectID": "src/axolotl/integrations/LICENSE.html", + "href": "src/axolotl/integrations/LICENSE.html", + "title": "Axolotl", + "section": "", + "text": "AXOLOTL COMMUNITY LICENSE AGREEMENT\nThis Axolotl Community License Agreement (“Agreement”) is entered into by and between Axolotl AI Corp. (“Axolotl”) and any individual or entity (“Licensee”) who wishes to use the Software (as defined below) in accordance with the terms and conditions set forth in this Agreement.\n\nDefinitions 1.1 “Licensee” refers to any individual or entity who has obtained a copy of the Software under this Agreement. 1.2 “Plugin Integration” means independent integration software modules which may or may not be offered by Axolotl, which may be licensed separately by their respective authors and/or licensors. 1.3 “Software” refers to the specific sub-directory of the Axolotl, Inc. software located at https://github.com/axolotl-ai-cloud/axolotl/tree/main/src/axolotl/integrations and its subdirectories which permits Plugin Integrations to integrate with the Axolotl service.\nGrant of License 2.1 Axolotl hereby grants Licensee a worldwide, non-exclusive, royalty-free, license to use, copy, modify, merge, publish, distribute, sublicense, and/or otherwise exploit the Software, subject to the following conditions: - Licensee must comply with all the terms and conditions of this Agreement. - Licensee must include the original copyright notice and disclaimer of warranty in all copies or substantial portions of the Software. 2.2 Licensee may use the Software for any lawful purpose, except as restricted in Section 3.\nRestrictions 3.1 Licensee shall not use the Software for any activity that constitutes a commercial activity of offering for free or for sale any services, platform, or equivalent to third parties for the purposes of allowing such third parties to fine-tune artificial intelligence models. 3.2 Licensee shall not: - Use the Software for any illegal or unauthorized purpose. - Reverse engineer, decompile, or disassemble the Software. - Remove or modify any copyright, trademark, or other proprietary notices contained in the Software. - Use the Software in a way that could damage, disable, overburden, or impair the functionality of the Software or interfere with any third-party use of the Software. 3.3 Axolotl reserves the right to restrict certain Plugin Integrations for use with the Software. To the extent Licensee integrates a permitted, applicable Plugin Integration with the Software, Licensee shall comply with any additional terms and conditions imposed by the licensors of such Plugin Integration for use of such Plugin Integrations. Licensee shall contact Axolotl if it has questions about whether its use of the Software falls beyond the scope of this Agreement.\nIntellectual Property Rights 4.1 Axolotl and its contributors retain all intellectual property rights in and to the Software. Licensee acknowledges that this Agreement does not transfer any ownership rights or intellectual property rights to Licensee.\nDisclaimer of Warranty 5.1 THE SOFTWARE IS PROVIDED “AS IS,” WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES, OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT, OR OTHERWISE, ARISING FROM, OUT OF, OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.\nTermination 6.1 Axolotl may terminate this Agreement at any time if Licensee fails to comply with any of the terms and conditions set forth herein. Upon termination, Licensee shall cease all use of the Software and destroy any copies in its possession.\nGoverning Law 7.1 This Agreement shall be governed by and construed in accordance with the laws of the State of California, without regards to conflicts of laws provisions thereof.\nEntire Agreement 8.1 This Agreement constitutes the entire agreement between Axolotl and Licensee with respect to the subject matter hereof and supersedes all prior or contemporaneous understandings or agreements between the parties concerning the Software, whether written or oral. Axolotl may update the terms of this Agreement from time to time, and Licensee’s continued use of the Software after any such updates shall constitute acceptance of updated terms on a go-forward basis. Axolotl will use commercially reasonable efforts to provide Licensee notice of any material updates. By using the Software, Licensee acknowledges that it has read, understood, and agrees to be bound by the terms and conditions of this Agreement.\n\nThis Agreement was last updated on August 23, 2024." }, { "objectID": "index.html", @@ -542,88 +514,123 @@ ] }, { - "objectID": "src/axolotl/integrations/LICENSE.html", - "href": "src/axolotl/integrations/LICENSE.html", - "title": "Axolotl", + "objectID": "examples/colab-notebooks/colab-axolotl-example.html", + "href": "examples/colab-notebooks/colab-axolotl-example.html", + "title": "Example notebook for running Axolotl on google colab", "section": "", - "text": "AXOLOTL COMMUNITY LICENSE AGREEMENT\nThis Axolotl Community License Agreement (“Agreement”) is entered into by and between Axolotl AI Corp. (“Axolotl”) and any individual or entity (“Licensee”) who wishes to use the Software (as defined below) in accordance with the terms and conditions set forth in this Agreement.\n\nDefinitions 1.1 “Licensee” refers to any individual or entity who has obtained a copy of the Software under this Agreement. 1.2 “Plugin Integration” means independent integration software modules which may or may not be offered by Axolotl, which may be licensed separately by their respective authors and/or licensors. 1.3 “Software” refers to the specific sub-directory of the Axolotl, Inc. software located at https://github.com/axolotl-ai-cloud/axolotl/tree/main/src/axolotl/integrations and its subdirectories which permits Plugin Integrations to integrate with the Axolotl service.\nGrant of License 2.1 Axolotl hereby grants Licensee a worldwide, non-exclusive, royalty-free, license to use, copy, modify, merge, publish, distribute, sublicense, and/or otherwise exploit the Software, subject to the following conditions: - Licensee must comply with all the terms and conditions of this Agreement. - Licensee must include the original copyright notice and disclaimer of warranty in all copies or substantial portions of the Software. 2.2 Licensee may use the Software for any lawful purpose, except as restricted in Section 3.\nRestrictions 3.1 Licensee shall not use the Software for any activity that constitutes a commercial activity of offering for free or for sale any services, platform, or equivalent to third parties for the purposes of allowing such third parties to fine-tune artificial intelligence models. 3.2 Licensee shall not: - Use the Software for any illegal or unauthorized purpose. - Reverse engineer, decompile, or disassemble the Software. - Remove or modify any copyright, trademark, or other proprietary notices contained in the Software. - Use the Software in a way that could damage, disable, overburden, or impair the functionality of the Software or interfere with any third-party use of the Software. 3.3 Axolotl reserves the right to restrict certain Plugin Integrations for use with the Software. To the extent Licensee integrates a permitted, applicable Plugin Integration with the Software, Licensee shall comply with any additional terms and conditions imposed by the licensors of such Plugin Integration for use of such Plugin Integrations. Licensee shall contact Axolotl if it has questions about whether its use of the Software falls beyond the scope of this Agreement.\nIntellectual Property Rights 4.1 Axolotl and its contributors retain all intellectual property rights in and to the Software. Licensee acknowledges that this Agreement does not transfer any ownership rights or intellectual property rights to Licensee.\nDisclaimer of Warranty 5.1 THE SOFTWARE IS PROVIDED “AS IS,” WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES, OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT, OR OTHERWISE, ARISING FROM, OUT OF, OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.\nTermination 6.1 Axolotl may terminate this Agreement at any time if Licensee fails to comply with any of the terms and conditions set forth herein. Upon termination, Licensee shall cease all use of the Software and destroy any copies in its possession.\nGoverning Law 7.1 This Agreement shall be governed by and construed in accordance with the laws of the State of California, without regards to conflicts of laws provisions thereof.\nEntire Agreement 8.1 This Agreement constitutes the entire agreement between Axolotl and Licensee with respect to the subject matter hereof and supersedes all prior or contemporaneous understandings or agreements between the parties concerning the Software, whether written or oral. Axolotl may update the terms of this Agreement from time to time, and Licensee’s continued use of the Software after any such updates shall constitute acceptance of updated terms on a go-forward basis. Axolotl will use commercially reasonable efforts to provide Licensee notice of any material updates. By using the Software, Licensee acknowledges that it has read, understood, and agrees to be bound by the terms and conditions of this Agreement.\n\nThis Agreement was last updated on August 23, 2024." + "text": "import torch\n# Check so there is a gpu available, a T4(free tier) is enough to run this notebook\nassert (torch.cuda.is_available()==True)" }, { - "objectID": "FAQS.html", - "href": "FAQS.html", - "title": "FAQs", - "section": "", - "text": "FAQs\n\nCan you train StableLM with this? Yes, but only with a single GPU atm. Multi GPU support is coming soon! Just waiting on this PR\nWill this work with Deepspeed? That’s still a WIP, but setting export ACCELERATE_USE_DEEPSPEED=true should work in some cases\nError invalid argument at line 359 in file /workspace/bitsandbytes/csrc/pythonInterface.c /arrow/cpp/src/arrow/filesystem/s3fs.cc:2598: arrow::fs::FinalizeS3 was not called even though S3 was initialized. This could lead to a segmentation fault at exit. Try reinstalling bitsandbytes and transformers from source." + "objectID": "examples/colab-notebooks/colab-axolotl-example.html#install-axolotl-and-dependencies", + "href": "examples/colab-notebooks/colab-axolotl-example.html#install-axolotl-and-dependencies", + "title": "Example notebook for running Axolotl on google colab", + "section": "Install Axolotl and dependencies", + "text": "Install Axolotl and dependencies\n\n!pip install -e git+https://github.com/axolotl-ai-cloud/axolotl#egg=axolotl\n!pip install flash-attn==\"2.5.0\"\n!pip install deepspeed==\"0.13.1\"!pip install mlflow==\"2.13.0\"" }, { - "objectID": "TODO.html", - "href": "TODO.html", - "title": "todo list", - "section": "", - "text": "[] Validation of parameters for combinations that won’t work\n\n\n\n\nFSDP offload and gradient_checkpointing - https://github.com/pytorch/pytorch/issues/82203\nadamw_bnb_8bit doesn’t play well with FSDP offload" + "objectID": "examples/colab-notebooks/colab-axolotl-example.html#create-an-yaml-config-file", + "href": "examples/colab-notebooks/colab-axolotl-example.html#create-an-yaml-config-file", + "title": "Example notebook for running Axolotl on google colab", + "section": "Create an yaml config file", + "text": "Create an yaml config file\n\nimport yaml\n\n# Your YAML string\nyaml_string = \"\"\"\nbase_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T\nmodel_type: LlamaForCausalLM\ntokenizer_type: LlamaTokenizer\n\nload_in_8bit: false\nload_in_4bit: true\nstrict: false\n\ndatasets:\n - path: mhenrichsen/alpaca_2k_test\n type: alpaca\ndataset_prepared_path:\nval_set_size: 0.05\noutput_dir: ./outputs/qlora-out\n\nadapter: qlora\nlora_model_dir:\n\nsequence_len: 4096\nsample_packing: true\neval_sample_packing: false\npad_to_sequence_len: true\n\nlora_r: 32\nlora_alpha: 16\nlora_dropout: 0.05\nlora_target_modules:\nlora_target_linear: true\nlora_fan_in_fan_out:\n\nwandb_project:\nwandb_entity:\nwandb_watch:\nwandb_name:\nwandb_log_model:\n\ngradient_accumulation_steps: 4\nmicro_batch_size: 2\nnum_epochs: 4\noptimizer: paged_adamw_32bit\nlr_scheduler: cosine\nlearning_rate: 0.0002\n\ntrain_on_inputs: false\ngroup_by_length: false\nbf16: auto\nfp16:\ntf32: false\n\ngradient_checkpointing: true\nearly_stopping_patience:\nresume_from_checkpoint:\nlocal_rank:\nlogging_steps: 1\nxformers_attention:\nflash_attention: true\n\nwarmup_steps: 10\nevals_per_epoch: 4\nsaves_per_epoch: 1\ndebug:\ndeepspeed:\nweight_decay: 0.0\nfsdp:\nfsdp_config:\nspecial_tokens:\n\n\"\"\"\n\n# Convert the YAML string to a Python dictionary\nyaml_dict = yaml.safe_load(yaml_string)\n\n# Specify your file path\nfile_path = 'test_axolotl.yaml'\n\n# Write the YAML file\nwith open(file_path, 'w') as file:\n yaml.dump(yaml_dict, file)" }, { - "objectID": "TODO.html#things-that-are-known-not-to-work", - "href": "TODO.html#things-that-are-known-not-to-work", - "title": "todo list", - "section": "", - "text": "FSDP offload and gradient_checkpointing - https://github.com/pytorch/pytorch/issues/82203\nadamw_bnb_8bit doesn’t play well with FSDP offload" + "objectID": "examples/colab-notebooks/colab-axolotl-example.html#launch-the-training", + "href": "examples/colab-notebooks/colab-axolotl-example.html#launch-the-training", + "title": "Example notebook for running Axolotl on google colab", + "section": "Launch the training", + "text": "Launch the training\n\n# By using the ! the comand will be executed as a bash command\n!accelerate launch -m axolotl.cli.train /content/test_axolotl.yaml" }, { - "objectID": "docs/unsloth.html", - "href": "docs/unsloth.html", - "title": "Unsloth", + "objectID": "examples/colab-notebooks/colab-axolotl-example.html#play-with-inference", + "href": "examples/colab-notebooks/colab-axolotl-example.html#play-with-inference", + "title": "Example notebook for running Axolotl on google colab", + "section": "Play with inference", + "text": "Play with inference\n\n# By using the ! the comand will be executed as a bash command\n!accelerate launch -m axolotl.cli.inference /content/test_axolotl.yaml \\\n --qlora_model_dir=\"./qlora-out\" --gradio" + }, + { + "objectID": "docs/input_output.html", + "href": "docs/input_output.html", + "title": "Template-free prompt construction", "section": "", - "text": "Overview\nUnsloth provides hand-written optimized kernels for LLM finetuning that slightly improve speed and VRAM over standard industry baselines.\n\n\nInstallation\nThe following will install unsloth from source and downgrade xformers as unsloth is incompatible with the most up to date libraries.\npip install --no-deps \"unsloth @ git+https://github.com/unslothai/unsloth.git\"\npip install --no-deps --force-reinstall xformers==0.0.26.post1\n\n\nUsing unsloth w Axolotl\nAxolotl exposes a few configuration options to try out unsloth and get most of the performance gains.\nOur unsloth integration is currently limited to the following model architectures: - llama\nThese options are specific to LoRA finetuning and cannot be used for multi-GPU finetuning\nunsloth_lora_mlp: true\nunsloth_lora_qkv: true\nunsloth_lora_o: true\nThese options are composable and can be used with multi-gpu finetuning\nunsloth_cross_entropy_loss: true\nunsloth_rms_norm: true\nunsloth_rope: true\n\n\nLimitations\n\nSingle GPU only; e.g. no multi-gpu support\nNo deepspeed or FSDP support (requires multi-gpu)\nLoRA + QLoRA support only. No full fine tunes or fp8 support.\nLimited model architecture support. Llama, Phi, Gemma, Mistral only\nNo MoE support.", + "text": "Background\n\nMasking Inputs\nYou may not want prompt templates\nThe input_output format\n\nUsage\n\n1. Prepare Data\n2. Use type: input_output\n3. Check the prompts", "crumbs": [ "How-To Guides", - "Unsloth" + "Template-free prompt construction" ] }, { - "objectID": "docs/rlhf.html", - "href": "docs/rlhf.html", - "title": "RLHF (Beta)", - "section": "", - "text": "Overview\nReinforcement 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:\n\nProximal Policy Optimization (PPO) (not yet supported in axolotl)\nDirect Preference Optimization (DPO)\nIdentity Preference Optimization (IPO)\n\n\n\nRLHF using Axolotl\n\n[!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.\n\nThe 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\n\nDPO\nrl: dpo\ndatasets:\n - path: Intel/orca_dpo_pairs\n split: train\n type: chatml.intel\n - path: argilla/ultrafeedback-binarized-preferences\n split: train\n type: chatml.argilla\n\n\nIPO\nrl: ipo\n\n\nORPO\nPaper: https://arxiv.org/abs/2403.07691\nrl: orpo\norpo_alpha: 0.1\nremove_unused_columns: false\n\nchat_template: chatml\ndatasets:\n - path: argilla/ultrafeedback-binarized-preferences-cleaned\n type: chat_template.argilla\n\n\nUsing local dataset files\ndatasets:\n - ds_type: json\n data_files:\n - orca_rlhf.jsonl\n split: train\n type: chatml.intel\n\n\nTrl autounwrap for peft\nTrl 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.\n# load ref model when adapter training.\nrl_adapter_ref_model: true", + "objectID": "docs/input_output.html#background", + "href": "docs/input_output.html#background", + "title": "Template-free prompt construction", + "section": "Background", + "text": "Background\n\n\nMasking Inputs\nOne of the most popular features of axolotl is setting the following configuration value:\ntrain_on_inputs: false\nIf you declare a dataset formats such as alpaca or chatml, axolotl knows what is an input (i.e. human) vs. an output (i.e. the assistant) and masks the input labels so that your model can focus on predicting the outputs only.\n\n\n\nYou may not want prompt templates\nHowever, there are many situations where you don’t want to use one of these formats or templates. This is because they can:\n\nAdd unnecessary boilerplate to your prompts.\nCreate artifacts like special delimiters <|im_start|> that can quickly become footguns if you don’t include them correctly at inference time.\nEnforce a chat interface when you do not want one. Sometimes you just want to fine-tune a model to a very specific task and do NOT want multi-turn conversations, roles, etc.\nLimit you to only certain roles that the template allows.\n\n\n\n\nThe input_output format\nYou can construct your prompts without a template by using the input_output format, by setting type: input_output in your configuration file like this:\nconfig.yml\ntrain_on_inputs: false # Mask segments of your data\ndatasets:\n - path: output.jsonl\n type: input_output # use template free prompt construction\nUnlike type: completion, which is also template-free, type: input_output allows you to mask segments of your text. More details on how this works are described below.", "crumbs": [ "How-To Guides", - "RLHF (Beta)" + "Template-free prompt construction" ] }, { - "objectID": "docs/mac.html", - "href": "docs/mac.html", - "title": "Mac M-series", - "section": "", - "text": "Currently Axolotl on Mac is partially usable, many of the dependencies of Axolotl including Pytorch do not support MPS or have incomplete support.\nCurrent support:\n\nSupport for all models\nFull training of models\nLoRA training\nSample packing\nFP16 and BF16 (awaiting AMP support for MPS in Pytorch)\nTri-dao’s flash-attn (until it is supported use spd_attention as an alternative)\nxformers\nbitsandbytes (meaning no 4/8 bits loading and bnb optimizers)\nqlora\nDeepSpeed\n\nUntested: - FSDP", + "objectID": "docs/input_output.html#usage", + "href": "docs/input_output.html#usage", + "title": "Template-free prompt construction", + "section": "Usage", + "text": "Usage\nThis is how you can use the input_output format:\n\n\n1. Prepare Data\nTo use the input_output format, collect your data in the following format into a jsonl file (below is the first row from the file output.jsonl` pretty printed):\n$ head -n1 output.jsonl | python -m json.tool\n\n{\n \"segments\": [\n {\n \"label\": true,\n \"text\": \"<s>Hello\\n\"\n },\n {\n \"label\": true,\n \"text\": \"hi there!. \"\n },\n {\n \"label\": false,\n \"text\": \"goodbye \"\n },\n {\n \"label\": true,\n \"text\": \"farewell</s>\"\n }\n ]\n}\n\nSet label:false when you want to mask a segment of text so that the model isn’t trained on it. Some things to keep in mind:\n\n[!IMPORTANT] 1. EOS, BOS, spaces, newlines etc. are entirely up to you. Axolotl concatenates all the segments as-is. The tokenizer doesn’t add anything additional. Notice how I added spaces, newlines, <s> (BOS), and </s> (EOS) myself. 2. Make sure you check the materialized output to validate that the prompt is getting assembled how you like.\n\n\n\n\n2. Use type: input_output\nLet’s materialize data with our output.jsonl file by setting type: input_output in our axolotl config:\n# training_config.yaml\nbase_model: mistralai/Mistral-7B-v0.1\ndata_seed: 49\nseed: 49\n\ndatasets:\n - path: output.jsonl\n type: input_output\nval_set_size: 0.1\n\nsequence_len: 896\nsample_packing: false\n\nmicro_batch_size: 2\ngradient_accumulation_steps: 3\neval_batch_size: 2\nnum_epochs: 1\nlearning_rate: 0.0002\n\ntrain_on_inputs: false\nspecial_tokens:\n bos_token: \"<s>\"\n eos_token: \"</s>\"\n unk_token: \"<unk>\"\nYou can use the following command to materialize your data. The --debug flag will print the tokens, along with the labels so you can verify that the correct items are being ignored:\n$ python -m axolotl.cli.preprocess training_config.yaml --debug\n\n...\n[2024-03-05 23:36:46,969] [INFO] [axolotl.check_example_labels:35] [PID:607731] [RANK:0] <s>(1, 1) Hello(22557, 22557)\n(13, 13) hi(12014, 12014) there(736, 736) !(28808, 28808) .(28723, 28723) (28705, 28705) good(-100, 1179) bye(-100, 17664) (-100, 28705) fare(19111, 19111) well(5458, 5458) </s>(2, 2)\nThe format is decoded_token(label, token_id), for example, <s>(1, 1) means that the token is <s>, the label is 1 and the token_id is 1. When the label is -100 then that token is ignored for training.\n\n\n\n3. Check the prompts\nHere is another way to check the materialized output:\nfrom transformers import AutoTokenizer\nfrom datasets import load_from_disk\nimport yaml\n\ndirectory = !ls last_run_prepared/\nwith open('training_config.yaml', 'r') as f:\n cfg = yaml.safe_load(f)\nmodel_id = cfg['base_model']\ntok = AutoTokenizer.from_pretrained(model_id)\nds = load_from_disk(f'last_run_prepared/{directory[0]}/')\n>>> row = ds[0]\n>>> print(tok.decode(row['input_ids']))\n<s> Hello\n hi there!. goodbye farewell</s>\nWe can check that the right tokens are ignored by comparing the labels to each token:\nimport pandas as pd\npd.DataFrame([{'token': tok.decode(i), 'label': l, 'id':i} for i,l in\n zip(row['input_ids'], row['labels'])])\n\n\n\ntoken\nlabel\nid\n\n\n\n\n0\n<s>\n1\n\n\n1\nHello\n22557\n\n\n2\n\\n\n13\n\n\n3\nhi\n12014\n\n\n4\nthere\n736\n\n\n5\n!\n28808\n\n\n6\n.\n28723\n\n\n7\n\n28705\n\n\n8\ngood\n-100\n\n\n9\nbye\n-100\n\n\n10\n\n-100\n\n\n11\nfare\n19111\n\n\n12\nwell\n5458\n\n\n13\n</s>\n2\n\n\n\nIf we look at the input data, the above table seems correct! (The jsonl version is repeated below for reference):\n$ head -n1 output.jsonl | python -m json.tool\n\n{\n \"segments\": [\n {\n \"label\": true,\n \"text\": \"<s>Hello\\n\"\n },\n {\n \"label\": true,\n \"text\": \"hi there!. \"\n },\n {\n \"label\": false,\n \"text\": \"goodbye \"\n },\n {\n \"label\": true,\n \"text\": \"farewell</s>\"\n }\n ]\n}", "crumbs": [ "How-To Guides", - "Mac M-series" + "Template-free prompt construction" ] }, { - "objectID": "docs/multi-node.html", - "href": "docs/multi-node.html", - "title": "Multi Node", + "objectID": "docs/amd_hpc.html", + "href": "docs/amd_hpc.html", + "title": "Training with AMD GPUs on HPC Systems", "section": "", - "text": "You will need to create a configuration for accelerate, either by using accelerate config and follow the instructions or you can use one of the preset below:\n~/.cache/huggingface/accelerate/default_config.yaml\nConfigure your model to use FSDP with for example:", + "text": "This guide provides step-by-step instructions for installing and configuring Axolotl on a High-Performance Computing (HPC) environment equipped with AMD GPUs.", "crumbs": [ "How-To Guides", - "Multi Node" + "Training with AMD GPUs on HPC Systems" ] }, { - "objectID": "docs/multi-node.html#machine-configuration", - "href": "docs/multi-node.html#machine-configuration", - "title": "Multi Node", - "section": "Machine configuration", - "text": "Machine configuration\nOn each machine you need a copy of Axolotl, we suggest using the same commit to ensure compatibility.\nYou will also need to have the same configuration file for your model on each machine.\nOn the main machine only, make sure the port you set as main_process_port is open in TCP and reachable by other machines.\nAll you have to do now is launch using accelerate as you would usually do on each machine and voila, the processes will start once you have launched accelerate on every machine.", + "objectID": "docs/amd_hpc.html#setup", + "href": "docs/amd_hpc.html#setup", + "title": "Training with AMD GPUs on HPC Systems", + "section": "Setup", + "text": "Setup\n\n1. Install Python\nWe recommend using Miniforge, a minimal conda-based Python distribution:\ncurl -L -O \"https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh\"\nbash Miniforge3-$(uname)-$(uname -m).sh\n\n\n2. Configure Python Environment\nAdd Python to your PATH and ensure it’s available at login:\necho 'export PATH=~/miniforge3/bin:$PATH' >> ~/.bashrc\necho 'if [ -f ~/.bashrc ]; then . ~/.bashrc; fi' >> ~/.bash_profile\n\n\n3. Load AMD GPU Software\nLoad the ROCm module:\nmodule load rocm/5.7.1\nNote: The specific module name and version may vary depending on your HPC system. Consult your system documentation for the correct module name.\n\n\n4. Install PyTorch\nInstall PyTorch with ROCm support:\npip install -U torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.7 --force-reinstall\n\n\n5. Install Flash Attention\nClone and install the Flash Attention repository:\ngit clone --recursive https://github.com/ROCmSoftwarePlatform/flash-attention.git\nexport GPU_ARCHS=\"gfx90a\"\ncd flash-attention\nexport PYTHON_SITE_PACKAGES=$(python -c 'import site; print(site.getsitepackages()[0])')\npatch \"${PYTHON_SITE_PACKAGES}/torch/utils/hipify/hipify_python.py\" hipify_patch.patch\npip install .\n\n\n6. Install Axolotl\nClone and install Axolotl:\ngit clone https://github.com/axolotl-ai-cloud/axolotl\ncd axolotl\npip install packaging ninja\npip install -e .\n\n\n7. Apply xformers Workaround\nxformers appears to be incompatible with ROCm. Apply the following workarounds: - Edit $HOME/packages/axolotl/src/axolotl/monkeypatch/llama_attn_hijack_flash.py modifying the code to always return False for SwiGLU availability from xformers. - Edit $HOME/miniforge3/lib/python3.10/site-packages/xformers/ops/swiglu_op.py replacing the “SwiGLU” function with a pass statement.\n\n\n8. Prepare Job Submission Script\nCreate a script for job submission using your HPC’s particular software (e.g. Slurm, PBS). Include necessary environment setup and the command to run Axolotl training. If the compute node(s) do(es) not have internet access, it is recommended to include\nexport TRANSFORMERS_OFFLINE=1\nexport HF_DATASETS_OFFLINE=1\n\n\n9. Download Base Model\nDownload a base model using the Hugging Face CLI:\nhuggingface-cli download meta-llama/Meta-Llama-3.1-8B --local-dir ~/hfdata/llama3.1-8B\n\n\n10. Create Axolotl Configuration\nCreate an Axolotl configuration file (YAML format) tailored to your specific training requirements and dataset. Use FSDP for multi-node training.\nNote: Deepspeed did not work at the time of testing. However, if anyone managed to get it working, please let us know.\n\n\n11. Preprocess Data\nRun preprocessing on the login node:\nCUDA_VISIBLE_DEVICES=\"\" python -m axolotl.cli.preprocess /path/to/your/config.yaml\n\n\n12. Train\nYou are now ready to submit your previously prepared job script. 🚂", "crumbs": [ "How-To Guides", - "Multi Node" + "Training with AMD GPUs on HPC Systems" ] }, + { + "objectID": "docs/config.html", + "href": "docs/config.html", + "title": "Config options", + "section": "", + "text": "# This is the huggingface model that contains *.pt, *.safetensors, or *.bin files\n# This can also be a relative path to a model on disk\nbase_model: ./llama-7b-hf\n# You can specify an ignore pattern if the model repo contains more than 1 model type (*.pt, etc)\nbase_model_ignore_patterns:\n# If the base_model repo on hf hub doesn't include configuration .json files,\n# You can set that here, or leave this empty to default to base_model\nbase_model_config: ./llama-7b-hf\n# You can specify to choose a specific model revision from huggingface hub\nrevision_of_model:\n# Optional tokenizer configuration path in case you want to use a different tokenizer\n# than the one defined in the base model\ntokenizer_config:\n# If you want to specify the type of model to load, AutoModelForCausalLM is a good choice too\nmodel_type: AutoModelForCausalLM\n# Corresponding tokenizer for the model AutoTokenizer is a good choice\ntokenizer_type: AutoTokenizer\n# Trust remote code for untrusted source\ntrust_remote_code:\n# use_fast option for tokenizer loading from_pretrained, default to True\ntokenizer_use_fast:\n# Whether to use the legacy tokenizer setting, defaults to True\ntokenizer_legacy:\n# Resize the model embeddings when new tokens are added to multiples of 32\n# This is reported to improve training speed on some models\nresize_token_embeddings_to_32x:\n\n# (Internal use only)\n# Used to identify which the model is based on\nis_falcon_derived_model:\nis_llama_derived_model:\nis_qwen_derived_model:\n# Please note that if you set this to true, `padding_side` will be set to \"left\" by default\nis_mistral_derived_model:\n\n# optional overrides to the base model configuration\noverrides_of_model_config:\n # RoPE Scaling https://github.com/huggingface/transformers/pull/24653\n rope_scaling:\n type: # linear | dynamic\n factor: # float\n\n# optional overrides to the bnb 4bit quantization configuration\n# https://huggingface.co/docs/transformers/main/main_classes/quantization#transformers.BitsAndBytesConfig\nbnb_config_kwargs:\n # These are default values\n llm_int8_has_fp16_weight: false\n bnb_4bit_quant_type: nf4\n bnb_4bit_use_double_quant: true\n\n\n# Whether you are training a 4-bit GPTQ quantized model\ngptq: true\n\n# This will attempt to quantize the model down to 8 bits and use adam 8 bit optimizer\nload_in_8bit: true\n# Use bitsandbytes 4 bit\nload_in_4bit:\n\n# Use CUDA bf16\nbf16: true # bool or 'full' for `bf16_full_eval`. require >=ampere\n# Use CUDA fp16\nfp16: true\n# Use CUDA tf32\ntf32: true # require >=ampere\n\n# No AMP (automatic mixed precision)\nbfloat16: true # require >=ampere\nfloat16: true\n\n# Limit the memory for all available GPUs to this amount (if an integer, expressed in gigabytes); default: unset\ngpu_memory_limit: 20GiB\n# Do the LoRA/PEFT loading on CPU -- this is required if the base model is so large it takes up most or all of the available GPU VRAM, e.g. during a model and LoRA merge\nlora_on_cpu: true\n\n# A list of one or more datasets to finetune the model with\ndatasets:\n # HuggingFace dataset repo | s3://,gs:// path | \"json\" for local dataset, make sure to fill data_files\n - path: vicgalle/alpaca-gpt4\n # The type of prompt to use for training. [alpaca, sharegpt, gpteacher, oasst, reflection]\n type: alpaca # format | format:<prompt_style> (chat/instruct) | <prompt_strategies>.load_<load_fn>\n ds_type: # Optional[str] (json|arrow|parquet|text|csv) defines the datatype when path is a file\n data_files: # Optional[str] path to source data files\n shards: # Optional[int] number of shards to split data into\n name: # Optional[str] name of dataset configuration to load\n train_on_split: train # Optional[str] name of dataset split to load from\n\n # Optional[str] fastchat conversation type, only used with type: sharegpt\n conversation: # Options (see Conversation 'name'): https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py\n field_human: # Optional[str]. Human key to use for conversation.\n field_model: # Optional[str]. Assistant key to use for conversation.\n # Add additional keys from your dataset as input or output roles\n roles:\n input: # Optional[List[str]]. These will be masked based on train_on_input\n output: # Optional[List[str]].\n\n # Custom user instruction prompt\n - path: repo\n type:\n # The below are defaults. only set what's needed if you use a different column name.\n system_prompt: \"\"\n system_format: \"{system}\"\n field_system: system\n field_instruction: instruction\n field_input: input\n field_output: output\n\n # Customizable to be single line or multi-line\n # Use {instruction}/{input} as key to be replaced\n # 'format' can include {input}\n format: |-\n User: {instruction} {input}\n Assistant:\n # 'no_input_format' cannot include {input}\n no_input_format: \"{instruction} \"\n\n # For `completion` datsets only, uses the provided field instead of `text` column\n field:\n\n# If false, the datasets will not be shuffled and will keep their original order in `datasets`.\n# The same applies to the `test_datasets` option and the `pretraining_dataset` option. Default is true.\nshuffle_merged_datasets: true\n\n# A list of one or more datasets to eval the model with.\n# You can use either test_datasets, or val_set_size, but not both.\ntest_datasets:\n - path: /workspace/data/eval.jsonl\n ds_type: json\n # You need to specify a split. For \"json\" datasets the default split is called \"train\".\n split: train\n type: completion\n data_files:\n - /workspace/data/eval.jsonl\n\n# use RL training: 'dpo', 'ipo', 'kto'\nrl:\n\n# Saves the desired chat template to the tokenizer_config.json for easier inferencing\n# Currently supports chatml and inst (mistral/mixtral)\nchat_template: chatml\n# Changes the default system message\ndefault_system_message: You are a helpful assistant. Please give a long and detailed answer. # Currently only supports chatml.\n# Axolotl attempts to save the dataset as an arrow after packing the data together so\n# subsequent training attempts load faster, relative path\ndataset_prepared_path: data/last_run_prepared\n# Push prepared dataset to hub\npush_dataset_to_hub: # repo path\n# The maximum number of processes to use while preprocessing your input dataset. This defaults to `os.cpu_count()`\n# if not set.\ndataset_processes: # defaults to os.cpu_count() if not set\n# Keep dataset in memory while preprocessing\n# Only needed if cached dataset is taking too much storage\ndataset_keep_in_memory:\n# push checkpoints to hub\nhub_model_id: # private repo path to push finetuned model\n# how to push checkpoints to hub\n# https://huggingface.co/docs/transformers/v4.31.0/en/main_classes/trainer#transformers.TrainingArguments.hub_strategy\nhub_strategy:\n# Whether to use hf `use_auth_token` for loading datasets. Useful for fetching private datasets\n# Required to be true when used in combination with `push_dataset_to_hub`\nhf_use_auth_token: # boolean\n# How much of the dataset to set aside as evaluation. 1 = 100%, 0.50 = 50%, etc. 0 for no eval.\nval_set_size: 0.04\n# Num shards for whole dataset\ndataset_shard_num:\n# Index of shard to use for whole dataset\ndataset_shard_idx:\n\n# The maximum length of an input to train with, this should typically be less than 2048\n# as most models have a token/context limit of 2048\nsequence_len: 2048\n# Pad inputs so each step uses constant sized buffers\n# This will reduce memory fragmentation and may prevent OOMs, by re-using memory more efficiently\npad_to_sequence_len:\n# Use efficient multi-packing with block diagonal attention and per sequence position_ids. Recommend set to 'true'\nsample_packing:\n# Set to 'false' if getting errors during eval with sample_packing on.\neval_sample_packing:\n# You can set these packing optimizations AFTER starting a training at least once.\n# The trainer will provide recommended values for these values.\nsample_packing_eff_est:\ntotal_num_tokens:\n# Increasing the following values helps with packing, but usually only slightly (<%1.)\n# The number of samples packed at a time.\nsample_packing_group_size: 100000\n# The number of samples which can be packed into one sequence. Increase if using a large sequence_len with many short samples.\nsample_packing_bin_size: 200\n\n# Passed through to transformers when loading the model when launched without accelerate\n# Use `sequential` when training w/ model parallelism to limit memory\ndevice_map:\n# Defines the max memory usage per gpu on the system. Passed through to transformers when loading the model.\nmax_memory:\n\n# If you want to use 'lora' or 'qlora' or leave blank to train all parameters in original model\nadapter: lora\n# If you already have a lora model trained that you want to load, put that here.\n# This means after training, if you want to test the model, you should set this to the value of `output_dir`.\n# Note that if you merge an adapter to the base model, a new subdirectory `merged` will be created under the `output_dir`.\nlora_model_dir:\n\n# LoRA hyperparameters\n# For more details about the following options, see:\n# https://www.anyscale.com/blog/fine-tuning-llms-lora-or-full-parameter-an-in-depth-analysis-with-llama-2\nlora_r: 8\nlora_alpha: 16\nlora_dropout: 0.05\nlora_target_modules:\n - q_proj\n - v_proj\n# - k_proj\n# - o_proj\n# - gate_proj\n# - down_proj\n# - up_proj\nlora_target_linear: # If true, will target all linear modules\npeft_layers_to_transform: # The layer indices to transform, otherwise, apply to all layers\n\n# If you added new tokens to the tokenizer, you may need to save some LoRA modules because they need to know the new tokens.\n# For LLaMA and Mistral, you need to save `embed_tokens` and `lm_head`. It may vary for other models.\n# `embed_tokens` converts tokens to embeddings, and `lm_head` converts embeddings to token probabilities.\n# https://github.com/huggingface/peft/issues/334#issuecomment-1561727994\nlora_modules_to_save:\n# - embed_tokens\n# - lm_head\n\nlora_fan_in_fan_out: false\n\n# LoRA+ hyperparameters\n# For more details about the following options, see:\n# https://arxiv.org/abs/2402.12354 and `src/axolotl/core/train_builder.py`\nloraplus_lr_ratio: # loraplus learning rate ratio lr_B / lr_A. Recommended value is 2^4.\nloraplus_lr_embedding: # loraplus learning rate for lora embedding layers. Default value is 1e-6.\n\npeft:\n # Configuration options for loftq initialization for LoRA\n # https://huggingface.co/docs/peft/developer_guides/quantization#loftq-initialization\n loftq_config:\n loftq_bits: # typically 4 bits\n\n# ReLoRA configuration\n# Must use either 'lora' or 'qlora' adapter, and does not support fsdp or deepspeed\nrelora_steps: # Number of steps per ReLoRA restart\nrelora_warmup_steps: # Number of per-restart warmup steps\nrelora_anneal_steps: # Number of anneal steps for each relora cycle\nrelora_prune_ratio: # threshold for optimizer magnitude when pruning\nrelora_cpu_offload: # True to perform lora weight merges on cpu during restarts, for modest gpu memory savings\n\n# wandb configuration if you're using it\n# Make sure your `WANDB_API_KEY` environment variable is set (recommended) or you login to wandb with `wandb login`.\nwandb_mode: # \"offline\" to save run metadata locally and not sync to the server, \"disabled\" to turn off wandb\nwandb_project: # Your wandb project name\nwandb_entity: # A wandb Team name if using a Team\nwandb_watch:\nwandb_name: # Set the name of your wandb run\nwandb_run_id: # Set the ID of your wandb run\nwandb_log_model: # \"checkpoint\" to log model to wandb Artifacts every `save_steps` or \"end\" to log only at the end of training\n\n# mlflow configuration if you're using it\nmlflow_tracking_uri: # URI to mlflow\nmlflow_experiment_name: # Your experiment name\nhf_mlflow_log_artifacts: # set to true to copy each saved checkpoint on each save to mlflow artifact registry\n\n# Where to save the full-finetuned model to\noutput_dir: ./completed-model\n\n# Whether to use torch.compile and which backend to use\ntorch_compile: # bool\ntorch_compile_backend: # Optional[str]\n\n# Training hyperparameters\n\n# If greater than 1, backpropagation will be skipped and the gradients will be accumulated for the given number of steps.\ngradient_accumulation_steps: 1\n# The number of samples to include in each batch. This is the number of samples sent to each GPU.\n# Batch size per gpu = micro_batch_size * gradient_accumulation_steps\nmicro_batch_size: 2\neval_batch_size:\nnum_epochs: 4\nwarmup_steps: 100 # cannot use with warmup_ratio\nwarmup_ratio: 0.05 # cannot use with warmup_steps\nlearning_rate: 0.00003\nlr_quadratic_warmup:\nlogging_steps:\neval_steps: # Leave empty to eval at each epoch, integers for every N steps. decimal for fraction of total steps\nevals_per_epoch: # number of times per epoch to run evals, mutually exclusive with eval_steps\nsave_strategy: # Set to `\"no\"` to skip checkpoint saves\nsave_steps: # Leave empty to save at each epoch\nsaves_per_epoch: # number of times per epoch to save a checkpoint, mutually exclusive with save_steps\nsave_total_limit: # Checkpoints saved at a time\n# Maximum number of iterations to train for. It precedes num_epochs which means that\n# if both are set, num_epochs will not be guaranteed.\n# e.g., when 1 epoch is 1000 steps => `num_epochs: 2` and `max_steps: 100` will train for 100 steps\nmax_steps:\n\neval_table_size: # Approximate number of predictions sent to wandb depending on batch size. Enabled above 0. Default is 0\neval_max_new_tokens: # Total number of tokens generated for predictions sent to wandb. Default is 128\neval_causal_lm_metrics: # HF evaluate metrics used during evaluation. Default is [\"sacrebleu\", \"comet\", \"ter\", chrf]\n\nloss_watchdog_threshold: # High loss value, indicating the learning has broken down (a good estimate is ~2 times the loss at the start of training)\nloss_watchdog_patience: # Number of high-loss steps in a row before the trainer aborts (default: 3)\n\n# Save model as safetensors (require safetensors package)\nsave_safetensors:\n\n# Whether to mask out or include the human's prompt from the training labels\ntrain_on_inputs: false\n# Group similarly sized data to minimize padding.\n# May be slower to start, as it must download and sort the entire dataset.\n# Note that training loss may have an oscillating pattern with this enabled.\ngroup_by_length: false\n\n# Whether to use gradient checkpointing https://huggingface.co/docs/transformers/v4.18.0/en/performance#gradient-checkpointing\ngradient_checkpointing: false\n# additional kwargs to pass to the trainer for gradient checkpointing\n# gradient_checkpointing_kwargs:\n# use_reentrant: true\n\n# Stop training after this many evaluation losses have increased in a row\n# https://huggingface.co/transformers/v4.2.2/_modules/transformers/trainer_callback.html#EarlyStoppingCallback\nearly_stopping_patience: 3\n\n# Specify a scheduler and kwargs to use with the optimizer\nlr_scheduler: # 'one_cycle' | 'log_sweep' | empty for cosine\nlr_scheduler_kwargs:\ncosine_min_lr_ratio: # decay lr to some percentage of the peak lr, e.g. cosine_min_lr_ratio=0.1 for 10% of peak lr\ncosine_constant_lr_ratio: # freeze lr at some percentage of the step, e.g. cosine_constant_lr_ratio=0.8 means start cosine_min_lr at 80% of training step (https://arxiv.org/pdf/2308.04014.pdf)\n\n# For one_cycle optim\nlr_div_factor: # Learning rate div factor\n\n# Specify optimizer\n# Valid values are driven by the Transformers OptimizerNames class, see:\n# https://github.com/huggingface/transformers/blob/95b374952dc27d8511541d6f5a4e22c9ec11fb24/src/transformers/training_args.py#L134\n#\n# Note that not all optimizers may be available in your environment, ex: 'adamw_anyprecision' is part of\n# torchdistx, 'adamw_bnb_8bit' is part of bnb.optim.Adam8bit, etc. When in doubt, it is recommended to start with the optimizer used\n# in the examples/ for your model and fine-tuning use case.\n#\n# Valid values for 'optimizer' include:\n# - adamw_hf\n# - adamw_torch\n# - adamw_torch_fused\n# - adamw_torch_xla\n# - adamw_apex_fused\n# - adafactor\n# - adamw_anyprecision\n# - sgd\n# - adagrad\n# - adamw_bnb_8bit\n# - lion_8bit\n# - lion_32bit\n# - paged_adamw_32bit\n# - paged_adamw_8bit\n# - paged_lion_32bit\n# - paged_lion_8bit\n# - galore_adamw\n# - galore_adamw_8bit\n# - galore_adafactor\n# - galore_adamw_layerwise\n# - galore_adamw_8bit_layerwise\n# - galore_adafactor_layerwise\noptimizer:\n# Dictionary of arguments to pass to the optimizer\noptim_args:\n# For Galore Optimizers the following optim_args are available\n# rank: # type: int\n# update_proj_gap # type: int\n# scale # type: float\n# proj_type: # type: str, default = std\n\n# The target modules to optimize, i.e. the module names that you would like to train, right now this is used only for GaLore algorithm\noptim_target_modules:\n# - self_attn # for llama\n# - mlp\n\n# Specify weight decay\nweight_decay:\n# adamw hyperparams\nadam_beta1:\nadam_beta2:\nadam_epsilon:\n# Gradient clipping max norm\nmax_grad_norm:\n\n# Augmentation techniques\n# NEFT https://arxiv.org/abs/2310.05914, set this to a number (paper default is 5) to add noise to embeddings\n# currently only supported on Llama and Mistral\nneftune_noise_alpha:\n\n# Whether to bettertransformers\nflash_optimum:\n# Whether to use xformers attention patch https://github.com/facebookresearch/xformers:\nxformers_attention:\n# Whether to use flash attention patch https://github.com/Dao-AILab/flash-attention:\nflash_attention:\nflash_attn_cross_entropy: # Whether to use flash-attention cross entropy implementation - advanced use only\nflash_attn_rms_norm: # Whether to use flash-attention rms norm implementation - advanced use only\nflash_attn_fuse_qkv: # Whether to fuse QKV into a single operation\nflash_attn_fuse_mlp: # Whether to fuse part of the MLP into a single operation\n# Whether to use scaled-dot-product attention\n# https://pytorch.org/docs/stable/generated/torch.nn.functional.scaled_dot_product_attention.html\nsdp_attention:\n# Shifted-sparse attention (only llama) - https://arxiv.org/pdf/2309.12307.pdf\ns2_attention:\n# Resume from a specific checkpoint dir\nresume_from_checkpoint:\n# If resume_from_checkpoint isn't set and you simply want it to start where it left off.\n# Be careful with this being turned on between different models.\nauto_resume_from_checkpoints: false\n\n# Don't mess with this, it's here for accelerate and torchrun\nlocal_rank:\n\n# Add or change special tokens.\n# If you add tokens here, you don't need to add them to the `tokens` list.\nspecial_tokens:\n # bos_token: \"<s>\"\n # eos_token: \"</s>\"\n # unk_token: \"<unk>\"\n # pad_token: \"[PAD]\"\n\n# Add extra tokens.\ntokens:\n\n# FSDP\nfsdp:\nfsdp_config:\n\n# Deepspeed config path. e.g., deepspeed_configs/zero3.json\ndeepspeed:\n\n# Advanced DDP Arguments\nddp_timeout:\nddp_bucket_cap_mb:\nddp_broadcast_buffers:\n\n# Path to torch distx for optim 'adamw_anyprecision'\ntorchdistx_path:\n\n# Set to HF dataset for type: 'completion' for streaming instead of pre-tokenize\npretraining_dataset:\n\n# Debug mode\ndebug:\n\n# Seed\nseed:\n\n# Allow overwrite yml config using from cli\nstrict:", + "crumbs": [ + "Reference", + "Config options" + ] + }, + { + "objectID": "docs/faq.html", + "href": "docs/faq.html", + "title": "FAQ", + "section": "", + "text": "Q: The trainer stopped and hasn’t progressed in several minutes.\n\nA: Usually an issue with the GPUs communicating with each other. See the NCCL doc\n\nQ: Exitcode -9\n\nA: This usually happens when you run out of system RAM.\n\nQ: Exitcode -7 while using deepspeed\n\nA: Try upgrading deepspeed w: pip install -U deepspeed\n\nQ: AttributeError: ‘DummyOptim’ object has no attribute ‘step’\n\nA: You may be using deepspeed with single gpu. Please don’t set deepspeed: in yaml or cli.", + "crumbs": [ + "FAQ" + ] + }, + { + "objectID": "docs/multimodal.html", + "href": "docs/multimodal.html", + "title": "MultiModal / Vision Language Models (BETA)", + "section": "", + "text": "MultiModal / Vision Language Models (BETA)\n\nSupported Models\n\nMllama, i.e. llama with vision models\n\n\n\nUsage\nCurrently multimodal support is limited and doesn’t have full feature parity. To finetune a multimodal Llama w/ LoRA, you’ll need to use the following in YAML in combination with the rest of the required hyperparams.\nbase_model: alpindale/Llama-3.2-11B-Vision-Instruct\nprocessor_type: AutoProcessor\nskip_prepare_dataset: true\n\nchat_template: llama3_2_vision\ndatasets:\n - path: HuggingFaceH4/llava-instruct-mix-vsft\n type: chat_template\n split: train[:1%]\n field_messages: messages\nremove_unused_columns: false\nsample_packing: false\n\n# only finetune the Language model, leave the vision model and vision tower frozen\nlora_target_modules: 'language_model.model.layers.[\\d]+.(mlp|cross_attn|self_attn).(up|down|gate|q|k|v|o)_proj'" + }, { "objectID": "docs/debugging.html", "href": "docs/debugging.html", diff --git a/sitemap.xml b/sitemap.xml index 0af7adba7..ab6de7969 100644 --- a/sitemap.xml +++ b/sitemap.xml @@ -2,106 +2,110 @@ https://axolotl-ai-cloud.github.io/axolotl/docs/nccl.html - 2024-09-30T17:56:28.928Z + 2024-10-03T01:03:01.031Z https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/pretraining.html - 2024-09-30T17:56:28.927Z + 2024-10-03T01:03:01.030Z https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/inst_tune.html - 2024-09-30T17:56:28.927Z + 2024-10-03T01:03:01.030Z https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/index.html - 2024-09-30T17:56:28.927Z + 2024-10-03T01:03:01.030Z https://axolotl-ai-cloud.github.io/axolotl/docs/batch_vs_grad.html - 2024-09-30T17:56:28.926Z + 2024-10-03T01:03:01.030Z https://axolotl-ai-cloud.github.io/axolotl/docs/torchao.html - 2024-09-30T17:56:28.928Z + 2024-10-03T01:03:01.032Z https://axolotl-ai-cloud.github.io/axolotl/docs/multipack.html - 2024-09-30T17:56:28.928Z - - - https://axolotl-ai-cloud.github.io/axolotl/docs/faq.html - 2024-09-30T17:56:28.927Z - - - https://axolotl-ai-cloud.github.io/axolotl/docs/config.html - 2024-09-30T17:56:28.926Z - - - https://axolotl-ai-cloud.github.io/axolotl/docs/amd_hpc.html - 2024-09-30T17:56:28.926Z - - - https://axolotl-ai-cloud.github.io/axolotl/docs/input_output.html - 2024-09-30T17:56:28.928Z - - - https://axolotl-ai-cloud.github.io/axolotl/examples/colab-notebooks/colab-axolotl-example.html - 2024-09-30T17:56:28.929Z - - - https://axolotl-ai-cloud.github.io/axolotl/index.html - 2024-09-30T17:56:28.939Z - - - https://axolotl-ai-cloud.github.io/axolotl/src/axolotl/integrations/LICENSE.html - 2024-09-30T17:56:28.941Z - - - https://axolotl-ai-cloud.github.io/axolotl/FAQS.html - 2024-09-30T17:56:28.925Z - - - https://axolotl-ai-cloud.github.io/axolotl/TODO.html - 2024-09-30T17:56:28.925Z - - - https://axolotl-ai-cloud.github.io/axolotl/docs/unsloth.html - 2024-09-30T17:56:28.928Z - - - https://axolotl-ai-cloud.github.io/axolotl/docs/rlhf.html - 2024-09-30T17:56:28.928Z - - - https://axolotl-ai-cloud.github.io/axolotl/docs/mac.html - 2024-09-30T17:56:28.928Z + 2024-10-03T01:03:01.031Z https://axolotl-ai-cloud.github.io/axolotl/docs/multi-node.html - 2024-09-30T17:56:28.928Z + 2024-10-03T01:03:01.031Z + + + https://axolotl-ai-cloud.github.io/axolotl/docs/mac.html + 2024-10-03T01:03:01.031Z + + + https://axolotl-ai-cloud.github.io/axolotl/docs/rlhf.html + 2024-10-03T01:03:01.031Z + + + https://axolotl-ai-cloud.github.io/axolotl/docs/unsloth.html + 2024-10-03T01:03:01.032Z + + + https://axolotl-ai-cloud.github.io/axolotl/TODO.html + 2024-10-03T01:03:01.029Z + + + https://axolotl-ai-cloud.github.io/axolotl/FAQS.html + 2024-10-03T01:03:01.028Z + + + https://axolotl-ai-cloud.github.io/axolotl/src/axolotl/integrations/LICENSE.html + 2024-10-03T01:03:01.044Z + + + https://axolotl-ai-cloud.github.io/axolotl/index.html + 2024-10-03T01:03:01.043Z + + + https://axolotl-ai-cloud.github.io/axolotl/examples/colab-notebooks/colab-axolotl-example.html + 2024-10-03T01:03:01.032Z + + + https://axolotl-ai-cloud.github.io/axolotl/docs/input_output.html + 2024-10-03T01:03:01.031Z + + + https://axolotl-ai-cloud.github.io/axolotl/docs/amd_hpc.html + 2024-10-03T01:03:01.030Z + + + https://axolotl-ai-cloud.github.io/axolotl/docs/config.html + 2024-10-03T01:03:01.030Z + + + https://axolotl-ai-cloud.github.io/axolotl/docs/faq.html + 2024-10-03T01:03:01.030Z + + + https://axolotl-ai-cloud.github.io/axolotl/docs/multimodal.html + 2024-10-03T01:03:01.031Z https://axolotl-ai-cloud.github.io/axolotl/docs/debugging.html - 2024-09-30T17:56:28.927Z + 2024-10-03T01:03:01.030Z https://axolotl-ai-cloud.github.io/axolotl/docs/dataset_preprocessing.html - 2024-09-30T17:56:28.927Z + 2024-10-03T01:03:01.030Z https://axolotl-ai-cloud.github.io/axolotl/docs/fsdp_qlora.html - 2024-09-30T17:56:28.927Z + 2024-10-03T01:03:01.030Z https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/template_free.html - 2024-09-30T17:56:28.927Z + 2024-10-03T01:03:01.030Z https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/conversation.html - 2024-09-30T17:56:28.926Z + 2024-10-03T01:03:01.030Z https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/tokenized.html - 2024-09-30T17:56:28.927Z + 2024-10-03T01:03:01.030Z