|
--- |
|
tags: |
|
- instruct |
|
- pytorch |
|
- finance |
|
- stock market |
|
- candlesticks |
|
- FinGPT |
|
- option trading |
|
- future stock prediction |
|
- trends prediction |
|
- Enterprise LLM |
|
- Enterprise |
|
- Enterprise ready |
|
- Banks |
|
- Wealth Management |
|
model-index: |
|
- name: Palmyra-Fin-70B-32k |
|
results: [] |
|
license: other |
|
license_name: writer-open-model-license |
|
license_link: https://writer.com/legal/open-model-license/ |
|
extra_gated_prompt: >- |
|
By clicking "Agree", you agree to the [License |
|
Agreement](https://writer.com/legal/open-model-license/) and acknowledge |
|
Writer's [Privacy Policy](https://writer.com/legal/acceptable-use/). |
|
extra_gated_fields: |
|
Name: text |
|
Email: text |
|
Organization or Affiliation: text |
|
Receive email updates and promotions on Writer products, services, and research?: |
|
type: select |
|
options: |
|
- 'Yes' |
|
- 'No' |
|
I acknowledge that this model is for non-commercial use only unless I acquire a separate license from Writer: checkbox |
|
language: |
|
- en |
|
widget: |
|
- example_title: Palmyra-Fin-70B-32k |
|
messages: |
|
- role: system |
|
content: >- |
|
You are an expert language model specialized in the financial services |
|
domain. Your responses should be detailed, accurate, and tailored to the |
|
needs of financial professionals. Consider ethical implications, long-term |
|
impacts, and current global financial trends in your advice. |
|
- role: user |
|
content: >- |
|
As the CFO of a technology company, we are considering issuing new equity |
|
to fund our expansion into international markets. Could you provide a |
|
brief analysis of the advantages and disadvantages of issuing new equity, |
|
especially in the context of maintaining our current share price and |
|
investor relations? |
|
output: |
|
text: >- |
|
As the CFO of a technology company, considering issuing new equity to fund |
|
expansion into international markets is a strategic decision that requires |
|
a thorough analysis of its advantages and disadvantages. Issuing new |
|
equity can be an effective way to raise capital, but its crucial to weigh |
|
the benefits against the potential dilution of existing shareholders |
|
ownership and its impact on your share price. Heres a balanced view of the |
|
key advantages and disadvantages to help you make an informed |
|
decision:\n\nAdvantages:\n\n1. **Capital Raising**: Issuing new equity is |
|
an effective way to raise capital for expansion into international |
|
markets. This capital can be used to establish a presence in new markets, |
|
hire local teams, and develop marketing campaigns tailored to those |
|
regions.\n\n2. **Diversified Shareholder Base**: Issuing new equity can |
|
attract new investors, potentially including international investors. This |
|
can help broaden your shareholder base, which may improve liquidity and |
|
provide greater market stability.\n\n3. **Enhanced Credibility**: Issuing |
|
new equity can be perceived by the market as a vote of confidence in your |
|
company's growth prospects, enhancing your credibility with existing and |
|
potential investors.\n\n4. **Strategic Partnerships**: Issuing new equity |
|
can facilitate strategic partnerships or collaborations. For example, |
|
issuing shares to a strategic partner in a new market can help you |
|
establish a local presence and gain access to their customer base.\n\n5. |
|
**Employee Incentives**: Issuing new equity can be used to incentivize |
|
employees, especially when vesting is tied to specific performance metrics |
|
or time-based vesting schedules. This can help attract and retain top |
|
talent.\n\nDisadvantages:\n\n1. **Dilution of Ownership**: Issuing new |
|
equity dilutes the ownership percentage of existing shareholders. This |
|
dilution can cause a decrease in the share price of your company's |
|
stock.\n\n2. **Dilution of Earnings**: Issuing new equity dilutes earnings |
|
per share, which could cause a decrease in your company's share |
|
price.\n\n3. **Increased Shareholder Scrutiny**: Issuing new equity may |
|
attract new shareholders who will demand greater transparency and scrutiny |
|
of your operations. This could increase legal and financial reporting |
|
expenses, make some activities more time-consuming, and may also place |
|
undue strain on your personnel.\n\n4. **Issuance Costs and Expenses**: |
|
Issuing new equity involves significant legal, accounting, and other |
|
expenses. These costs could include, among other things, registration |
|
fees, filing fees, printing fees, and the compensation of any |
|
underwriters. The total amount of these costs may be substantial.\n\n5. |
|
**Market Volatility**: Issuing new equity can cause market volatility in |
|
the price of your company's stock. The sale of these shares, or the |
|
perception that sales could occur, could cause the price of your company's |
|
stock to fall. This could also make it more difficult for you to sell |
|
equity or debt securities in the future at a time or price that you deem |
|
reasonable or appropriate.\n\n6. **Perception of Weakness**: The sale of |
|
additional equity securities may lead to a perception by the market that |
|
your company's management believes its existing capital resources will be |
|
insufficient for its operations. This perception could cause your |
|
company's stock price to decline.\n\n7. **Restrictions on Dividends**: If |
|
you are incorporated in a state that limits your ability to pay dividends |
|
on your common stock while you have preferred stock outstanding, then the |
|
rights of the holders of your common stock may be subject to limitations |
|
on the declaration of dividends.\n\n8. **Anti-Takeover Effects**: Issuing |
|
new equity may have anti-takeover effects. For example, if your company |
|
has a staggered board of directors, shares underlying new equity could be |
|
issued by your board of directors in one or more rights plans or similar |
|
plans. The rights will entitle all shareholders of record on a certain |
|
date to purchase shares of preferred stock upon the occurrence of certain |
|
events, such as a person acquiring a certain percentage of your companys |
|
outstanding voting stock. The rights will expire prior to the |
|
commencement of any business combination with a person who beneficially |
|
owns 20% or more of your companys outstanding voting stock. The rights may |
|
have the effect of entrenching your companys management team and may |
|
deprive you of the opportunity to sell your company to potential acquirers |
|
at a premium over prevailing prices. This potential inability to obtain a |
|
control premium could reduce the price of your company's stock.\n\n9. |
|
**Impact on Employee Stock Options**: Issuing new equity may cause your |
|
companys employee stock options to become less valuable. This could |
|
adversely affect your companys ability to attract and retain |
|
employees.\n\n10. **Alternative Financing Options**: Depending on your |
|
companys stage of development and market conditions, it may be possible to |
|
obtain needed capital through alternative financing sources, such as a |
|
second public offering, a private placement of securities, or loans from |
|
banks or related or unrelated parties. These financing sources may have |
|
terms favorable to your company compared to those obtained through an |
|
issuance of new equity.\n\nIn the context of maintaining your current |
|
share price and investor relations, it's essential to consider the |
|
following:\n\n1. **Communicate Effectively**: It is crucial to |
|
communicate the reasons behind issuing new equity clearly and |
|
transparently to your existing shareholders. This can help manage |
|
expectations and reduce potential negative perceptions.\n\n2. **Share |
|
Price Impact**: Analyze the potential dilution of earnings per share and |
|
the ownership percentage of existing shareholders. Consider the impact of |
|
issuing new equity on your companys share price in the short and long |
|
term.\n\n3. **Investor Relations**: Engage with your existing shareholders |
|
to understand their concerns and address them proactively. This can help |
|
maintain a positive relationship with your investor base.\n\n4. **Equity |
|
Incentive Plans**: Ensure that your equity incentive plans are aligned |
|
with your companys long-term goals and that vesting schedules are |
|
structured to incentivize employees appropriately.\n\n5. **Alternative |
|
Financing Options**: Carefully evaluate alternative financing options and |
|
consider the potential impact of each on your share price and investor |
|
relations.\n\nIn conclusion, issuing new equity to fund expansion into |
|
international markets can be a viable option, but it is essential to |
|
carefully weigh the advantages and disadvantages and consider alternative |
|
financing options. Effective communication with your existing |
|
shareholders is critical to managing expectations and maintaining a |
|
positive relationship with your investor base. |
|
base_model: |
|
- Writer/palmyra-4-oasis |
|
--- |
|
|
|
|
|
|
|
<div align="center"> |
|
|
|
<h1>Palmyra-Fin, a powerful LLM designed for Finance</h1> |
|
<img src="img/Palmyra Finance wordmark.svg" width=420px /> |
|
</div> |
|
|
|
## Model Description |
|
|
|
- Developed by: Writer |
|
- Language(s) (NLP): English |
|
- Context window: 32,768 tokens |
|
- Parameters: 70 billion |
|
- Finetuned from model: Palmyra-X-003 |
|
- License: [Writer open model license](https://writer.com/legal/acceptable-use/) |
|
|
|
## Model Details |
|
|
|
Palmyra-Fin-70B-32K is a model built by Writer specifically to meet the needs of the financial industry. It is a leading LLM on financial benchmarks, outperforming other large language models in various financial tasks and evaluations. |
|
|
|
### Resources and Technical Documentation: |
|
|
|
+ [Writer Blog](https://writer.com/blog/palmyra-med-fin-models/) |
|
+ [Writer Developer Website](https://dev.writer.com/home/models) |
|
+ [Writer AI Studio](https://writer.com/product/ai-studio/) |
|
+ [Palmyra Model API](https://dev.writer.com/api-guides/chat-completion) |
|
|
|
|
|
## Specialized for Financial Applications |
|
|
|
Palmyra-Fin-70B-32K is meticulously designed to meet the unique linguistic and knowledge demands of the finance and economics sectors. It has been fine-tuned on an extensive collection of high-quality financial data, ensuring it can comprehend and generate text with precise domain-specific accuracy and fluency. |
|
|
|
Our system integrates a specialized internal finance dataset and a well-crafted fine-tuning recipe, making it highly adept at handling the specific needs of this field. Key components of our training pipeline include: |
|
|
|
- Specialized Dataset: Utilizing a proprietary internal finance dataset to enhance the model's performance. |
|
- Fine-tuning approach: Custom financial instruction dataset (Writer in-house build) |
|
|
|
|
|
## Intended Use |
|
|
|
### Intended Use Cases |
|
Palmyra-Fin-70B-32K is intended for use in English for financial analysis, market trend prediction, risk assessment, financial report generation, and automated financial advice. It excels at answering questions from long financial documents, making it ideal for in-depth financial research and analysis. |
|
|
|
### Out-of-scope |
|
Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by [Writer's Acceptable Use Policy](https://writer.com/legal/acceptable-use/) and [the Writer open model license](https://writer.com/legal/acceptable-use/). Use in languages other than English. |
|
|
|
**Note:** Users should be aware that while the model is highly capable, it should not be used as the sole basis for making significant financial decisions. |
|
|
|
|
|
### Use with transformers |
|
|
|
```python |
|
import torch |
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
|
model_id = "Writer/Palmyra-Fin-70B-32K" |
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model_id) |
|
|
|
model = AutoModelForCausalLM.from_pretrained( |
|
model_id, |
|
torch_dtype=torch.float16, |
|
device_map="auto", |
|
attn_implementation="flash_attention_2", |
|
) |
|
|
|
messages = [ |
|
{ |
|
"role": "system", |
|
"content": "You are a highly knowledgeable and experienced expert in the financial sector, possessing extensive knowledge and practical expertise in financial analysis, markets, investments, and economic principles.", |
|
}, |
|
{ |
|
"role": "user", |
|
"content": "Can you explain how central banks printing more money (quantitative easing) affects the stock market and how investors might react to it?", |
|
}, |
|
] |
|
|
|
input_ids = tokenizer.apply_chat_template( |
|
messages, tokenize=True, add_generation_prompt=True, return_tensors="pt" |
|
) |
|
|
|
gen_conf = { |
|
"max_new_tokens": 1024, |
|
"eos_token_id": tokenizer.eos_token_id, |
|
"temperature": 0.0, |
|
"top_p": 0.9, |
|
} |
|
|
|
with torch.inference_mode(): |
|
output_id = model.generate(input_ids, **gen_conf) |
|
|
|
output_text = tokenizer.decode(output_id[0][input_ids.shape[1] :]) |
|
|
|
print(output_text) |
|
``` |
|
|
|
|
|
## Evaluation Results |
|
|
|
Palmyra-Fin-70B-32K outperforms other models on internal finance evaluations, achieving state-of-the-art results across various financial datasets. Its strong performance in tasks like financial document analysis, market trend prediction, and risk assessment underscores its effective grasp of financial knowledge. |
|
|
|
Key Performance Metrics: |
|
|
|
- 100% accuracy on needle-in-haystack tasks |
|
- Superior performance on internal finance evaluations compared to other models |
|
|
|
Palmyra-Fin-70B-32K achieves 100% accuracy on needle-in-haystack tasks across its entire 32,768 token context window, demonstrating exceptional capability in precise information extraction from extensive financial documents. |
|
|
|
![palmyra-fin-70b-32k-nih](img/Palmyra-Fin-70b-32k-on-NIH.png) |
|
|
|
## CFA Level III Test |
|
The new Palmyra-Fin-70B model passed the CFA Level III test with a 73%. It's the first model to pass this exam, often called "the world's hardest exam." The average passing score is 60%, and less than half of all test takers pass. This is a big improvement from other models like GPT-4, which scored 33% on the exam. |
|
|
|
|
|
## Long-Fin-Eval Performance: |
|
|
|
To further assess the model's capabilities, we developed and conducted an evaluation using long-fin-eval, an internally created benchmark designed to simulate real-world financial use cases. This evaluation consists of samples containing long documents paired with high-quality question-answer sets. The model's task is to generate responses based on the provided document and question, with the output evaluated by GPT-4 Turbo. |
|
|
|
The long-fin-eval methodology assesses both the model's information retrieval capabilities and its ability to engage in extended dialogue on complex financial topics. This approach provides insight into the model's capacity to process and synthesize information from lengthy financial documents while maintaining coherent and contextually appropriate conversational output. |
|
|
|
In this evaluation, Palmyra-Fin-70B-32K showed superior performance compared to both open-source and proprietary benchmark models. These results indicate the model's effectiveness in addressing real-world financial applications that require both comprehensive understanding of extensive documents and the ability to articulate nuanced insights. |
|
|
|
|
|
| model name | long-fin-eval | |
|
| -------------------- | -------------- | |
|
| Palmyra-Fin-70B-32K | 9.04 | |
|
| Claude 3.5 Sonnet | 9.02 | |
|
| Qwen-2 70B instruct | 8.9 | |
|
| gpt-4o | 8.72 | |
|
| palmyra-fin-56b | 8.23 | |
|
| mixtral-8x7b | 7.57 | |
|
|
|
![palmyra-fin-70b-32k-performance](img/performance.png) |
|
|
|
|
|
## Financial Use Cases |
|
|
|
Palmyra-Fin-70B-32K excels in analyzing and summarizing complex financial reports, market data, and economic indicators, extracting key information to generate concise, structured summaries. It helps enhance financial decision-making by performing advanced entity recognition, identifying key financial concepts such as market trends, economic indicators, and financial instruments from unstructured text. |
|
|
|
By leveraging its deep understanding of financial terminology, the model enhances information retrieval, data analysis, and knowledge discovery from financial reports, research articles, and other economic sources. These capabilities support applications like investment analysis, risk management, and financial research. |
|
|
|
## Bias, Risks, and Limitations |
|
|
|
Palmyra-Fin-70B-32K, despite leveraging high-quality data, may contain inaccuracies, biases, or misalignments and has not been rigorously evaluated in real-world financial settings. |
|
|
|
It is advised not to use the model for direct financial decision-making or professional financial advice without human oversight. Instead, its use should be confined to research and analysis by qualified individuals who understand its limitations. Palmyra-Fin-70B-32K should not replace professional financial judgment, and adapting it for critical financial use would require extensive additional work, including thorough testing, regulatory compliance, bias mitigation, and human oversight. Always consult a qualified financial professional for personal financial needs. |
|
|
|
|
|
## Citation and Related Information |
|
To cite this model: |
|
``` |
|
@misc{Palmyra-Fin-70B-32k, |
|
author = {Writer Engineering team}, |
|
title = {{Palmyra-Fin-70B-32k: a powerful LLM designed for Finance}}, |
|
howpublished = {\url{https://dev.writer.com}}, |
|
year = 2024, |
|
month = July |
|
} |
|
``` |
|
Contact [email protected] |