Edit model card

Llama_Passthrough

Llama_Passthrough is a merged LLM using the following models using mergekit:

Passthrough:

Passthrough is a merging method where the embeddings or representations from multiple sources are concatenated or combined without any transformation or alignment.

🧩 Configuration

slices:
  - sources:
    - model: Suchinthana/Sinhala-Translate-and-Dolly-Llama-7b
      layer_range: [0, 32]
  - sources:
    - model: austinm2151/Llama2-7b-Summarizer
      layer_range: [24, 32]
merge_method: passthrough
dtype: bfloat16
Downloads last month
2
Safetensors
Model size
8.36B params
Tensor type
BF16
·