--- inference: false license: other datasets: - bavest/fin-llama-dataset tags: - finance - llm - llama - trading ---
TheBlokeAI

Chat & support: my new Discord server

Want to contribute? TheBloke's Patreon page

# Bavest's Fin Llama 33B GGML These files are GGML format model files for [Bavest's Fin Llama 33B](https://huggingface.co/bavest/fin-llama-33b-merged). GGML files are for CPU + GPU inference using [llama.cpp](https://github.com/ggerganov/llama.cpp) and libraries and UIs which support this format, such as: * [text-generation-webui](https://github.com/oobabooga/text-generation-webui) * [KoboldCpp](https://github.com/LostRuins/koboldcpp) * [ParisNeo/GPT4All-UI](https://github.com/ParisNeo/gpt4all-ui) * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) * [ctransformers](https://github.com/marella/ctransformers) ## Repositories available * [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/fin-llama-33B-GPTQ) * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/fin-llama-33B-GGML) * [Unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/bavest/fin-llama-33b-merged) ## Prompt template Standard Alpaca prompting: ``` A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's question. ### Instruction: prompt ### Response: ``` or ``` A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's question. ### Instruction: prompt ### Input: ### Response: ``` ## Compatibility ### Original llama.cpp quant methods: `q4_0, q4_1, q5_0, q5_1, q8_0` I have quantized these 'original' quantisation methods using an older version of llama.cpp so that they remain compatible with llama.cpp as of May 19th, commit `2d5db48`. They should be compatible with all current UIs and libraries that use llama.cpp, such as those listed at the top of this README. ### New k-quant methods: `q2_K, q3_K_S, q3_K_M, q3_K_L, q4_K_S, q4_K_M, q5_K_S, q6_K` These new quantisation methods are only compatible with llama.cpp as of June 6th, commit `2d43387`. They will NOT be compatible with koboldcpp, text-generation-ui, and other UIs and libraries yet. Support is expected to come over the next few days. ## Explanation of the new k-quant methods The new methods available are: * GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw) * GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw. * GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw. * GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw * GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw * GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type. Refer to the Provided Files table below to see what files use which methods, and how. ## Provided files | Name | Quant method | Bits | Size | Max RAM required | Use case | | ---- | ---- | ---- | ---- | ---- | ----- | | fin-llama-33b.ggmlv3.q2_K.bin | q2_K | 2 | 13.60 GB | 16.10 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. | | fin-llama-33b.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 17.20 GB | 19.70 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K | | fin-llama-33b.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 15.64 GB | 18.14 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K | | fin-llama-33b.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 13.98 GB | 16.48 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors | | fin-llama-33b.ggmlv3.q4_0.bin | q4_0 | 4 | 18.30 GB | 20.80 GB | Original llama.cpp quant method, 4-bit. | | fin-llama-33b.ggmlv3.q4_1.bin | q4_1 | 4 | 20.33 GB | 22.83 GB | Original llama.cpp quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. | | fin-llama-33b.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 19.57 GB | 22.07 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K | | fin-llama-33b.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 18.30 GB | 20.80 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors | | fin-llama-33b.ggmlv3.q5_0.bin | q5_0 | 5 | 22.37 GB | 24.87 GB | Original llama.cpp quant method, 5-bit. Higher accuracy, higher resource usage and slower inference. | | fin-llama-33b.ggmlv3.q5_1.bin | q5_1 | 5 | 24.40 GB | 26.90 GB | Original llama.cpp quant method, 5-bit. Even higher accuracy, resource usage and slower inference. | | fin-llama-33b.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 23.02 GB | 25.52 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K | | fin-llama-33b.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 22.37 GB | 24.87 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors | | fin-llama-33b.ggmlv3.q6_K.bin | q6_K | 6 | 26.69 GB | 29.19 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors | | fin-llama-33b.ggmlv3.q8_0.bin | q8_0 | 8 | 34.56 GB | 37.06 GB | Original llama.cpp quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users. | **Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead. ## How to run in `llama.cpp` I use the following command line; adjust for your tastes and needs: ``` ./main -t 10 -ngl 32 -m fin-llama-33b.ggmlv3.q5_0.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "### Instruction: I want you to act as an accountant and come up with creative ways to manage finances. You'll need to consider budgeting, investment strategies and risk management when creating a financial plan for your client. In some cases, you may also need to provide advice on taxation laws and regulations in order to help them maximize their profits. My first suggestion request is “Create a financial plan for a small business that focuses on cost savings and long-term investments".\n### Response:" ``` Change `-t 10` to the number of physical CPU cores you have. For example if your system has 8 cores/16 threads, use `-t 8`. Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration. If you want to have a chat-style conversation, replace the `-p ` argument with `-i -ins` ## How to run in `text-generation-webui` Further instructions here: [text-generation-webui/docs/llama.cpp-models.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp-models.md). ## Discord For further support, and discussions on these models and AI in general, join us at: [TheBloke AI's Discord server](https://discord.gg/Jq4vkcDakD) ## Thanks, and how to contribute. Thanks to the [chirper.ai](https://chirper.ai) team! I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training. If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects. Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits. * Patreon: https://patreon.com/TheBlokeAI * Ko-Fi: https://ko-fi.com/TheBlokeAI **Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov. **Patreon special mentions**: Oscar Rangel, Eugene Pentland, Talal Aujan, Cory Kujawski, Luke, Asp the Wyvern, Ai Maven, Pyrater, Alps Aficionado, senxiiz, Willem Michiel, Junyu Yang, trip7s trip, Sebastain Graf, Joseph William Delisle, Lone Striker, Jonathan Leane, Johann-Peter Hartmann, David Flickinger, Spiking Neurons AB, Kevin Schuppel, Mano Prime, Dmitriy Samsonov, Sean Connelly, Nathan LeClaire, Alain Rossmann, Fen Risland, Derek Yates, Luke Pendergrass, Nikolai Manek, Khalefa Al-Ahmad, Artur Olbinski, John Detwiler, Ajan Kanaga, Imad Khwaja, Trenton Dambrowitz, Kalila, vamX, webtim, Illia Dulskyi. Thank you to all my generous patrons and donaters! # Original model card: Bavest's Fin Llama 33B # FIN-LLAMA > Efficient Finetuning of Quantized LLMs for Finance [Adapter Weights](https://huggingface.co/bavest/fin-llama-33b-merged) | [Dataset](https://huggingface.co/datasets/bavest/fin-llama-dataset) ## Installation To load models in 4bits with transformers and bitsandbytes, you have to install accelerate and transformers from source and make sure you have the latest version of the bitsandbytes library (0.39.0). ```bash pip3 install -r requirements.txt ``` ### Other dependencies If you want to finetune the model on a new instance. You could run the `setup.sh` to install the python and cuda package. ```bash bash scripts/setup.sh ``` ## Finetuning ```bash bash script/finetune.sh ``` ## Usage Quantization parameters are controlled from the `BitsandbytesConfig` - Loading in 4 bits is activated through `load_in_4bit` - The datatype used for the linear layer computations with `bnb_4bit_compute_dtype` - Nested quantization is activated through `bnb_4bit_use_double_quant` - The datatype used for qunatization is specified with `bnb_4bit_quant_type`. Note that there are two supported quantization datatypes `fp4` (four bit float) and `nf4` (normal four bit float). The latter is theoretically optimal for normally distributed weights and we recommend using `nf4`. ```python import torch from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig pretrained_model_name_or_path = "bavest/fin-llama-33b-merge" model = AutoModelForCausalLM.from_pretrained( pretrained_model_name_or_path=pretrained_model_name_or_path, load_in_4bit=True, device_map='auto', torch_dtype=torch.bfloat16, quantization_config=BitsAndBytesConfig( load_in_4bit=True, bnb_4bit_compute_dtype=torch.bfloat16, bnb_4bit_use_double_quant=True, bnb_4bit_quant_type='nf4' ), ) tokenizer = AutoTokenizer.from_pretrained(pretrained_model_name_or_path) question = "What is the market cap of apple?" input = "" # context if needed prompt = f""" A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's question. '### Instruction:\n{question}\n\n### Input:{input}\n""\n\n### Response: """ input_ids = tokenizer.encode(prompt, return_tensors="pt").to('cuda:0') with torch.no_grad(): generated_ids = model.generate( input_ids, do_sample=True, top_p=0.9, temperature=0.8, max_length=128 ) generated_text = tokenizer.decode( [el.item() for el in generated_ids[0]], skip_special_tokens=True ) ``` ## Dataset for FIN-LLAMA The dataset is released under bigscience-openrail-m. You can find the dataset used to train FIN-LLAMA models on HF at [bavest/fin-llama-dataset](https://huggingface.co/datasets/bavest/fin-llama-dataset). ## Known Issues and Limitations Here a list of known issues and bugs. If your issue is not reported here, please open a new issue and describe the problem. See [QLORA](https://github.com/artidoro/qlora) for any other limitations. 1. 4-bit inference is slow. Currently, our 4-bit inference implementation is not yet integrated with the 4-bit matrix multiplication 2. Currently, using `bnb_4bit_compute_type='fp16'` can lead to instabilities. 3. Make sure that `tokenizer.bos_token_id = 1` to avoid generation issues. ## Acknowledgements We also thank Meta for releasing the LLaMA models without which this work would not have been possible. This repo builds on the [Stanford Alpaca](https://github.com/tatsu-lab/stanford_alpaca) , [QLORA](https://github.com/artidoro/qlora), [Chinese-Guanaco](https://github.com/jianzhnie/Chinese-Guanaco/tree/main) and [LMSYS FastChat](https://github.com/lm-sys/FastChat) repos. ## License and Intended Use We release the resources associated with QLoRA finetuning in this repository under GLP3 license. In addition, we release the FIN-LLAMA model family for base LLaMA model sizes of 7B, 13B, 33B, and 65B. These models are intended for purposes in line with the LLaMA license and require access to the LLaMA models. ## Prompts ### Act as an Accountant > I want you to act as an accountant and come up with creative ways to manage finances. You'll need to consider budgeting, investment strategies and risk management when creating a financial plan for your client. In some cases, you may also need to provide advice on taxation laws and regulations in order to help them maximize their profits. My first suggestion request is “Create a financial plan for a small business that focuses on cost savings and long-term investments". ## Paged Optimizer You can access the paged optimizer with the argument --optim paged_adamw_32bit ## Cite ```tex @misc{Fin-LLAMA, author = {William Todt, Ramtin Babaei, Pedram Babaei}, title = {Fin-LLAMA: Efficient Finetuning of Quantized LLMs for Finance}, year = {2023}, publisher = {GitHub}, journal = {GitHub repository}, howpublished = {\url{https://github.com/Bavest/fin-llama}}, } ```