G

MythoMax 13B

Gryphe
Vision
4K
Context
$0.0600
Input /1M
$0.0600
Output /1M
N/A
Max Output
Run API Call Open in Chat

MythoMax 13B

Overview

Property Value
Model Name MythoMax 13B
Model ID gryphe/mythomax-l2-13b
Creator Gryphe
Created July 2, 2023
Base Model Llama 2 13B (fine-tune/merge)
Architecture Transformer (LLaMA 2)

Description

One of the highest performing and most popular fine-tunes of Llama 2 13B, with rich descriptions and roleplay. This is a merged model (#merge) that combines multiple fine-tuning approaches to achieve superior creative writing and roleplay capabilities.

MythoMax was created by combining the storytelling abilities of MythoLogic-L2 with the creative writing prowess of Chronos 13B, resulting in a model that excels at generating rich, descriptive narrative content while maintaining coherent character interactions.

Specifications

Specification Value
Context Window 4,000 tokens
Context Length 4,096 tokens
Parameters 13 billion
Input Modalities Text
Output Modalities Text
Instruction Type Alpaca
Quantization Available int4, int8, fp16

Pricing

Primary Provider (NextBit)

Type Cost (per million tokens)
Input $0.06
Output $0.06

Pricing may vary by provider and quantization level.

Capabilities

Capability Status
Chat Completions Supported
Text Completions Supported
Reasoning/Chain-of-Thought Not specialized
Function Calling Not supported
Vision/Image Input Not supported
Code Generation Limited

Supported Parameters

Parameter Supported Description
Structured Outputs Yes JSON schema-based output formatting
Response Format Yes Control output format (text, json_object)
Max Tokens Yes Limit response length
Temperature Yes Control randomness (0.0 - 2.0)
Top P Yes Nucleus sampling parameter
Stop Sequences Yes Custom stop strings
Frequency Penalty Yes Reduce repetition of tokens
Presence Penalty Yes Encourage topic diversity

Use Cases

Strengths

  • Creative Writing: Generates rich, descriptive prose with vivid imagery
  • Roleplay: Excellent character consistency and dialogue
  • Storytelling: Strong narrative structure and plot development
  • Fantasy/Fiction: Particularly effective for fantasy and fiction genres
  • Character Dialogue: Natural, engaging character conversations
  • Interactive fiction and text adventures
  • Roleplay chat applications
  • Creative writing assistance
  • Story generation and continuation
  • Character development and dialogue

Limitations

  • Limited to 4K context window
  • Not optimized for factual/knowledge tasks
  • No vision or multimodal capabilities
  • Not suitable for coding tasks
  • May require careful prompting for non-creative tasks
Model Relation Description
Llama 2 13B Base Original foundation model
MythoLogic-L2-13B Parent Logic and storytelling focused
Chronos-13B Parent Creative writing focused
MythoMax-L2-13B-Kimiko-v2 Variant Extended context version
Noromaid-13B Similar Another roleplay-focused Llama 2 fine-tune
Pygmalion-13B Similar Conversational AI fine-tune

Default Configuration

Setting Value
Default Stop Sequences ###, </s>
Recommended Temperature 0.7 - 1.0
Recommended Top P 0.9 - 0.95

Provider Information

Provider Model ID Quantization Adapter
NextBit mythomax:13b int4 OpenAIAdapter

Data Policy

Policy Status
Training Data Retention No
Prompt Retention No
Model Weights Available on Hugging Face

Model Origins

MythoMax 13B is a merge of:

  • MythoLogic-L2-13B: Focused on logical reasoning and storytelling
  • Chronos-13B: Specialized in creative writing and narrative generation

The merge was performed using the TIES merging technique to combine the strengths of both models while minimizing conflicts.

Prompt Format

MythoMax uses the Alpaca instruction format:

### Instruction:
{instruction}

### Input:
{input}

### Response:
{response}

For chat/roleplay scenarios:

### Instruction:
You are {character_name}, {character_description}. Respond to the following in character.

### Input:
{user_message}

### Response:

Performance Metrics

MythoMax 13B is community-rated highly for:

  • Roleplay Quality: Excellent
  • Creative Writing: Excellent
  • Narrative Consistency: Very Good
  • Character Voice: Very Good
  • Response Length/Detail: Good

Note: Specific benchmark scores vary by evaluation methodology. The model was optimized for creative tasks rather than traditional NLP benchmarks.

Integration Example

LangMart API

curl https://api.langmart.ai/v1/chat/completions \
  -H "Authorization: Bearer $LANGMART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gryphe/mythomax-l2-13b",
    "messages": [
      {
        "role": "system",
        "content": "You are a skilled fantasy storyteller. Write with rich descriptions and engaging dialogue."
      },
      {
        "role": "user",
        "content": "Write the opening scene of a story about a wandering knight who discovers an abandoned castle."
      }
    ],
    "temperature": 0.8,
    "max_tokens": 1024,
    "stop": ["###", "</s>"]
  }'

LangMart Integration

curl -X POST https://api.langmart.ai/v1/chat/completions \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gryphe/mythomax-l2-13b",
    "messages": [
      {"role": "user", "content": "Write a short fantasy scene."}
    ],
    "temperature": 0.8
  }'

External Resources

Last updated: December 2024