DeepSeek-V3/inference
Cristian Cezar Moisés af62bffeff
Update generate.py
Error Handling: Improved error handling for file operations and JSON loading.
Logging: Clearer logging messages for better debugging and monitoring.
Code Comments: Added more descriptive comments to enhance code readability.
Dynamic Sequence Management: Clarified the logic for managing prompt lengths and token generation.
Performance: Minor optimizations in code structure and logic flow for better performance.
Code Structure: Organized functions and constants for better readability and maintainability.
2025-01-27 23:26:23 -03:00
..
configs Release DeepSeek-V3 2024-12-26 19:01:57 +08:00
convert.py Update convert.py 2025-01-27 23:23:28 -03:00
fp8_cast_bf16.py Update fp8_cast_bf16.py 2025-01-27 23:24:46 -03:00
generate.py Update generate.py 2025-01-27 23:26:23 -03:00
kernel.py Update kernel.py 2025-01-27 23:19:26 -03:00
model.py Update model.py 2025-01-27 23:21:33 -03:00
requirements.txt Release DeepSeek-V3 2024-12-26 19:01:57 +08:00