RichardErkhov/avinash31d_-_phi-2-slerp-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

phi-2-slerp - GGUF

• Model creator: https://huggingface.co/avinash31d/
• Original model: https://huggingface.co/avinash31d/phi-2-slerp/

Name	Quant method	Size
phi-2-slerp.Q2_K.gguf	Q2_K	1.03GB
phi-2-slerp.IQ3_XS.gguf	IQ3_XS	1.12GB
phi-2-slerp.IQ3_S.gguf	IQ3_S	1.16GB
phi-2-slerp.Q3_K_S.gguf	Q3_K_S	1.16GB
phi-2-slerp.IQ3_M.gguf	IQ3_M	1.23GB
phi-2-slerp.Q3_K.gguf	Q3_K	1.33GB
phi-2-slerp.Q3_K_M.gguf	Q3_K_M	1.33GB
phi-2-slerp.Q3_K_L.gguf	Q3_K_L	1.47GB
phi-2-slerp.IQ4_XS.gguf	IQ4_XS	1.43GB
phi-2-slerp.Q4_0.gguf	Q4_0	1.49GB
phi-2-slerp.IQ4_NL.gguf	IQ4_NL	1.5GB
phi-2-slerp.Q4_K_S.gguf	Q4_K_S	1.51GB
phi-2-slerp.Q4_K.gguf	Q4_K	1.62GB
phi-2-slerp.Q4_K_M.gguf	Q4_K_M	1.62GB
phi-2-slerp.Q4_1.gguf	Q4_1	1.65GB
phi-2-slerp.Q5_0.gguf	Q5_0	1.8GB
phi-2-slerp.Q5_K_S.gguf	Q5_K_S	1.8GB
phi-2-slerp.Q5_K.gguf	Q5_K	1.87GB
phi-2-slerp.Q5_K_M.gguf	Q5_K_M	1.87GB
phi-2-slerp.Q5_1.gguf	Q5_1	1.95GB
phi-2-slerp.Q6_K.gguf	Q6_K	2.13GB
phi-2-slerp.Q8_0.gguf	Q8_0	2.75GB

Original model description:

license: mit tags: - merge - mergekit - lazymergekit - microsoft/phi-2 - rhysjones/phi-2-orange-v2 base_model: - microsoft/phi-2 - rhysjones/phi-2-orange-v2

phi-2-slerp

phi-2-slerp is a merge of the following models using LazyMergekit:

• microsoft/phi-2
• rhysjones/phi-2-orange-v2

🧩 Configuration

slices:
  - sources:
      - model: microsoft/phi-2
        layer_range: [0, 32]
      - model: rhysjones/phi-2-orange-v2
        layer_range: [0, 32]
merge_method: slerp
base_model: microsoft/phi-2
parameters:
  t:
    - filter: self_attn
      value: [0, 0.5, 0.3, 0.7, 1]
    - filter: mlp
      value: [1, 0.5, 0.7, 0.3, 0]
    - value: 0.5
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "avinash31d/phi-2-slerp"
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"])