oopere
/

FinChat-XS

@@ -1,199 +1,186 @@
 ---
 library_name: transformers
-tags: []
 ---
-# Model Card for Model ID
-<!-- Provide a quick summary of what the model is/does. -->
-## Model Details
-### Model Description
-<!-- Provide a longer summary of what this model is. -->
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
-- **Developed by:** [More Information Needed]
-- **Funded by [optional]:** [More Information Needed]
-- **Shared by [optional]:** [More Information Needed]
-- **Model type:** [More Information Needed]
-- **Language(s) (NLP):** [More Information Needed]
-- **License:** [More Information Needed]
-- **Finetuned from model [optional]:** [More Information Needed]
-### Model Sources [optional]
-<!-- Provide the basic links for the model. -->
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
-## Uses
-<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
-### Direct Use
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
-[More Information Needed]
-### Downstream Use [optional]
-<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
-[More Information Needed]
-### Out-of-Scope Use
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
-[More Information Needed]
-## Bias, Risks, and Limitations
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
-[More Information Needed]
-### Recommendations
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
-Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
-## How to Get Started with the Model
-Use the code below to get started with the model.
-[More Information Needed]
-## Training Details
-### Training Data
-<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
-[More Information Needed]
-### Training Procedure
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
-#### Preprocessing [optional]
-[More Information Needed]
-#### Training Hyperparameters
-- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
-#### Speeds, Sizes, Times [optional]
-<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
-[More Information Needed]
-## Evaluation
-<!-- This section describes the evaluation protocols and provides the results. -->
-### Testing Data, Factors & Metrics
-#### Testing Data
-<!-- This should link to a Dataset Card if possible. -->
-[More Information Needed]
-#### Factors
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
-[More Information Needed]
-#### Metrics
-<!-- These are the evaluation metrics being used, ideally with a description of why. -->
-[More Information Needed]
-### Results
-[More Information Needed]
-#### Summary
-## Model Examination [optional]
-<!-- Relevant interpretability work for the model goes here -->
-[More Information Needed]
-## Environmental Impact
-<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
-Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
-- **Hardware Type:** [More Information Needed]
-- **Hours used:** [More Information Needed]
-- **Cloud Provider:** [More Information Needed]
-- **Compute Region:** [More Information Needed]
-- **Carbon Emitted:** [More Information Needed]
-## Technical Specifications [optional]
-### Model Architecture and Objective
-[More Information Needed]
-### Compute Infrastructure
-[More Information Needed]
-#### Hardware
-[More Information Needed]
-#### Software
-[More Information Needed]
-## Citation [optional]
-<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
-**BibTeX:**
-[More Information Needed]
-**APA:**
-[More Information Needed]
-## Glossary [optional]
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
-[More Information Needed]
-## More Information [optional]
-[More Information Needed]
-## Model Card Authors [optional]
-[More Information Needed]
-## Model Card Contact
-[More Information Needed]

 ---
 library_name: transformers
+tags:
+- finance
+- chat
+license: apache-2.0
+datasets:
+- sujet-ai/Sujet-Finance-Instruct-177k
+language:
+- en
+base_model:
+- HuggingFaceTB/SmolLM2-360M-Instruct
 ---
+# FinChat-XS
+FinChat-XS is a lightweight financial domain language model designed to answer questions about finance, markets, investments, and economics in a conversational style.
+## Model Overview
+FinChat-XS is a fine-tuned version of [HuggingFaceTB/SmolLM2-360M-Instruct](https://huggingface.co/HuggingFaceTB/SmolLM2-360M-Instruct), optimized for financial domain conversations using LoRA (Low-Rank Adaptation). With only 360M parameters, it offers a balance between performance and efficiency, making it accessible for deployment on consumer hardware.
+The model combines professional financial knowledge with a conversational communication style, making it suitable for applications where users need expert financial information delivered in an approachable manner.
+## Repository & Resources
+- **Model weights**: Available in this repository
+- **Training code**: [GitHub Notebook](https://github.com/peremartra/FinLLMOpt/tree/main/FinChat-XS)
+- **Base model**: [HuggingFaceTB/SmolLM2-360M-Instruct](https://huggingface.co/HuggingFaceTB/SmolLM2-360M-Instruct)
+- **Training dataset**: [sujet-ai/Sujet-Finance-Instruct-177k](https://huggingface.co/datasets/sujet-ai/Sujet-Finance-Instruct-177k)
+## How the Model was Created
+FinChat-XS was developed through a focused fine-tuning process designed to enhance financial domain expertise while maintaining conversational abilities:
+1. **Base model selection**: Started with SmolLM2-360M-Instruct, a lightweight instruction-tuned language model
+2. **Dataset preparation**:
+   - Filtered the sujet-ai/Sujet-Finance-Instruct-177k dataset to focus on QA and conversational QA examples
+   - Applied length filtering to keep responses below 500 characters
+   - Augmented short conversational QA examples to improve conciseness
+3. **Fine-tuning approach**:
+   - Applied LoRA (Low-Rank Adaptation) to efficiently fine-tune the model
+   - Targeted key attention modules (q_proj, v_proj)
+   - Used rank r=4 and alpha=16
+   - Training configuration:
+     - Batch size: 2 (effective batch size 16 with gradient accumulation)
+     - Learning rate: 1.5e-4
+     - BF16 precision
+## Challenges
+The primary challenge encountered during the development of FinChat-XS was the lack of high-quality conversational datasets specifically focused on personal finance. While the Sujet-Finance-Instruct-177k dataset provided valuable financial QA examples, there remains a notable gap in naturalistic, multi-turn conversations about personal financial scenarios.
+## Why Use This Model?
+FinChat-XS offers several advantages for specific use cases:
+- **Efficient deployment**: At only 362MB, it can run on devices with limited resources.
+- **Financial domain knowledge**: Fine-tuned specifically on financial QA data
+- **Balanced communication style**: Combines professional financial knowledge with conversational delivery
+- **Low deployment cost**: Requires significantly less computational resources than larger models
+- **Customizable**: The LoRA adapter can be mixed with other adapters or further fine-tuned
+Ideal for:
+- Embedded financial assistants in mobile apps
+- Personal financial planning tools
+- Educational applications about finance and investing
+- Customer service automation for financial institutions
+- Quick deployment scenarios where larger models aren't practical
+## How to Use the Model
+### Basic Usage with Transformers
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+# Load model and tokenizer
+model_name = "oopere/FinChat-XS"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16)
+# Create a conversation
+messages = [
+    {"role": "user", "content": "What's the difference between stocks and bonds?"}
+]
+# Format the prompt using the chat template
+prompt = tokenizer.apply_chat_template(messages, tokenize=False)
+# Tokenize the prompt
+inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+# Generate a response
+outputs = model.generate(
+    **inputs,
+    max_new_tokens=256,
+    temperature=0.7,
+    top_p=0.9,
+    do_sample=True,
+    repetition_penalty=1.2
+)
+# Decode and print the response
+response = tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True)
+print(response)
+```
+### Optimized Inference with 8-bit Quantization
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
+import torch
+# Configure 8-bit quantization
+bnb_config = BitsAndBytesConfig(
+    load_in_8bit=True,
+    bnb_4bit_compute_dtype=torch.float16
+)
+# Load model with quantization
+model = AutoModelForCausalLM.from_pretrained(
+    "oopere/FinChat-XS",
+    quantization_config=bnb_config,
+    device_map="auto"
+)
+tokenizer = AutoTokenizer.from_pretrained("oopere/FinChat-XS")
+# Continue with the same usage pattern as above
+```
+### Using with LoRA Adapter Only
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import PeftModel, PeftConfig
+# Load base model
+base_model = AutoModelForCausalLM.from_pretrained("HuggingFaceTB/SmolLM2-360M-Instruct")
+tokenizer = AutoTokenizer.from_pretrained("HuggingFaceTB/SmolLM2-360M-Instruct")
+# Load LoRA adapter
+peft_model = PeftModel.from_pretrained(base_model, "oopere/qa-adapterFinChat-XS")
+# Continue with the same usage pattern as above
+```
+## Limitations & Considerations
+While FinChat-XS performs well in many financial conversation scenarios, users should be aware of these limitations:
+1. **Knowledge limitations**: The model's knowledge is limited to its training data and has a knowledge cutoff date from the base model (SmolLM2).
+2. **Size trade-offs**: As a 360M parameter model, it has less capacity than larger models (7B+) and may provide less nuanced or detailed responses on complex topics.
+3. **Financial advice disclaimer**: The model is not a certified financial advisor and should not be used for making investment decisions. Its responses should be considered educational, not professional financial advice.
+4. **Domain boundaries**: While focused on finance, the model may struggle with highly specialized financial topics or recent developments not covered in its training data.
+5. **Hallucination potential**: Like all language models, FinChat-XS may occasionally generate plausible-sounding but incorrect information, especially when asked about specific numerical data or complex financial details.
+6. **Style variations**: The model balances formal financial knowledge with a conversational style, which may not be appropriate for all professional contexts.
+7. **Regulatory compliance**: This model has not been specifically audited for compliance with financial regulations in various jurisdictions.
+## Citation
+If you use FinChat-XS in your research or applications, please consider citing it as:
+```
+@misc{oopere2025finchatxs,
+  author = {Martra, P.},
+  title = {FinChat-XS: A Lightweight Financial Domain Chat Language Model},
+  year = {2025},
+  publisher = {Hugging Face},
+  howpublished = {\url{https://huggingface.co/oopere/FinChat-XS}}
+}
+```
+## Acknowledgements
+- [HuggingFaceTB](https://huggingface.co/HuggingFaceTB) for creating the SmolLM2 model series
+- [Sujet AI](https://huggingface.co/sujet-ai) for their financial instruction dataset
+- [Hugging Face](https://huggingface.co/) for providing the infrastructure and tools for model development