lucataco/dolphin-2.1-mistral-7b 🔢📝 → 📝

▶️ 13.5K runs 📅 Oct 2023 ⚙️ Cog 0.8.5 🔗 GitHub 📄 Paper ⚖️ License
code-generation question-answering text-generation text-translation

About

Mistral-7B-v0.1 fine tuned for chat with the Dolphin dataset (an open-source implementation of Microsoft's Orca)

Example Output

Prompt:

"What is the best way to train a dolphin to obey me? Please answer step by step."

Output

Training a dolphin to obey you involves establishing trust, communication, and a strong foundation in basic behaviors. Here's a step-by-step guide to training a dolphin:

  1. Building trust: Spend time with the dolphin, observe its behavior, and let it get used to your presence. This will help in building trust, which is the cornerstone of any successful training relationship.

  2. Learning the dolphin's body language: Observe the dolphin's body language and learn to interpret its signals. This will help you understand when the dolphin is relaxed, excited, or fearful, and will allow you to respond appropriately.

  3. Using positive reinforcement: When teaching a behavior, start by rewarding the dolphin with food or attention as soon as it begins to respond correctly. This will help the dolphin associate the desired behavior with a positive outcome.

  4. Breaking down behaviors into smaller steps: Instead of trying to teach a complex behavior all at once, break it down into smaller, more manageable steps. This makes it easier for the dolphin to learn and understand each part of the behavior.

  5. Using clear and consistent signals: Use consistent hand signals, verbal commands, or other signals to communicate with the dolphin. This will help the dolphin understand what you want it to do and will make training more efficient.

  6. Practicing patience: Training a dolphin requires patience, as they may not learn a behavior as quickly as you would like. Be patient and give the dolphin time to learn and practice the behavior.

  7. Reinforcing learned behaviors: Once a behavior has been learned, continue to reinforce it periodically. This will help the dolphin maintain and strengthen its understanding of the behavior.

  8. Encouraging creativity and problem-solving: Dolphins are intelligent creatures that enjoy problem-solving and new challenges. Encourage the dolphin to think creatively and solve problems by presenting new situations or tasks for it to complete.

  9. Maintaining a strong bond: Continue to nurture the bond between you and the dolphin by spending time together, engaging in play, and communicating openly and honestly. This will help ensure a successful and lasting training relationship.

  10. Adapting to the dolphin's needs:

Performance Metrics

6.52s Prediction Time
7.88s Total Time
All Input Parameters 
{
  "top_k": 50,
  "top_p": 0.95,
  "prompt": "What is the best way to train a dolphin to obey me?  Please answer step by step.",
  "temperature": 0.8,
  "max_new_tokens": 512,
  "prompt_template": "<|im_start|>system\nyou are an expert dolphin trainer\n<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant\n",
  "presence_penalty": 0,
  "frequency_penalty": 0
}
Input Parameters
top_k Type: integerDefault: 50
The number of highest probability tokens to consider for generating the output. If > 0, only keep the top k tokens with highest probability (top-k filtering).
top_p Type: numberDefault: 0.95
A probability threshold for generating the output. If < 1.0, only keep the top tokens with cumulative probability >= top_p (nucleus filtering). Nucleus filtering is described in Holtzman et al. (http://arxiv.org/abs/1904.09751).
prompt (required) Type: string
temperature Type: numberDefault: 0.8
The value used to modulate the next token probabilities.
max_new_tokens Type: integerDefault: 512
The maximum number of tokens the model should generate as output.
prompt_template Type: stringDefault: <|im_start|>system You are Dolphin, a helpful AI assistant. <|im_end|> <|im_start|>user {prompt}<|im_end|> <|im_start|>assistant
The template used to format the prompt. The input prompt is inserted into the template using the `{prompt}` placeholder.
presence_penalty Type: numberDefault: 0
Presence penalty
frequency_penalty Type: numberDefault: 0
Frequency penalty
Output Schema

Output

Type: array • Items Type: string

Example Execution Logs 
Generated 512 tokens in 6.507347106933594 seconds.
Version Details
Version ID
7e6410619678c44898cd6077d5d79a64e392cf94125ddf710e76dec45af86755
Version Created
October 31, 2023
Run on Replicate →