deepseek-ai
/
deepseek-r1
A reasoning model trained with reinforcement learning, on par with OpenAI o1
Run replicate/deepseek-r1-fireworks-proxy-internal with an API
Input schema
top_pnumber
Top-p (nucleus) sampling
- Default
- 1
promptstring
Prompt
max_tokensinteger
The maximum number of tokens the model should generate as output.
- Default
- 20480
temperaturenumber
The value used to modulate the next token probabilities.
- Default
- 0.1
presence_penaltynumber
Presence penalty
frequency_penaltynumber
Frequency penalty
Output schema
- Type
- string[]