Model Type | |
Use Cases |
Areas: | research, commercial applications |
|
Primary Use Cases: | |
Limitations: | mostly trained on English data, may not generalize to other languages |
|
Considerations: | Develop guardrails and take appropriate precautions for any production use. |
|
|
Additional Notes | Falcon LLMs require PyTorch 2.0 for use with transformers. |
|
Supported Languages | English (high), French (medium) |
|
Training Details |
Data Sources: | Bai ze, GPT4All, GPTeacher, RefinedWeb-English |
|
Data Volume: | |
Context Length: | |
Hardware Used: | |
Model Architecture: | Causal decoder-only model with rotary positionnal embeddings, multiquery and FlashAttention attention, parallel attention/MLP with a single layer norm |
|
|
Responsible Ai Considerations |
Mitigation Strategies: | Develop guardrails and take appropriate precautions for any production use. |
|
|