L3-ColdBrew-Hades

L3-ColdBrew-Hades is a merge of the following models using LazyMergekit:

• SvalTek/L3-ColdBrew-Astrid
• SvalTek/L3-ColdBrew-Astrid
• SvalTek/L3-ColdBrew-Astrid
• SvalTek/L3-ColdBrew-Astrid
• SvalTek/L3-ColdBrew-Astrid

🧩 Configuration

# Use: mergekit-yaml --clone-tensors ./llama-3-attenuated.yaml ./llama-3-attenuated
# See: https://github.com/arcee-ai/mergekit/issues/198 for discussion/reasoning behind this idea.

# ---

# The scale factor to use, eg: solve x^2 = 1/2 --> x = 1/sqrt(2) ≈ 0.7071067812
const_tag: &scale_factor 0.7071067812 # 1/sqrt(2)

# The filter parameters of a scaled block. 
attenuate-env: &attenuated_env
  parameters:
    scale:
      - filter: q_proj
        value: *scale_factor
      - filter: k_proj
        value: *scale_factor
      - value: 1.0

# ---

slices:

  - sources:
    - model: SvalTek/L3-ColdBrew-Astrid
      layer_range: [0, 8] # The first 8 layers of Block 1 are not duplicated
  - sources:
    - model: SvalTek/L3-ColdBrew-Astrid
      layer_range: [8, 16] # The last 8 layers of Block 1 are are duplicated twice
      <<: *attenuated_env

  - sources:
    - model: SvalTek/L3-ColdBrew-Astrid
      layer_range: [8, 24] # All the layers of Block 2 are are duplicated twice
      <<: *attenuated_env

  - sources:
    - model: SvalTek/L3-ColdBrew-Astrid
      layer_range: [16, 24] # The first 8 layers of Block 3 are are duplicated twice
      <<: *attenuated_env
  - sources:
    - model: SvalTek/L3-ColdBrew-Astrid
      layer_range: [24, 32] # The last 8 layers of Block 3 are not duplicated

merge_method: passthrough
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "SvalTek/L3-ColdBrew-Hades"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])