Edit model card

Everyone else is using hidden gates with no prompts which I don't think works.

base_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
gate_mode: random
dtype: bfloat16
experts:
  - source_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
    positive_prompts: [""]
  - source_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
    positive_prompts: [""]
  - source_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
    positive_prompts: [""]
  - source_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
    positive_prompts: [""]
  - source_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
    positive_prompts: [""]
  - source_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
    positive_prompts: [""]
  - source_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
    positive_prompts: [""]
  - source_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
    positive_prompts: [""]
Downloads last month
2
Safetensors
Model size
6.43B params
Tensor type
BF16
·