You can deploy this model and start earning money today!

RichardErkhov/
abacaj_-_phi-2-super-gguf

Name: abacaj_-_phi-2-super-gguf
Availability: InStock

Views

No views yet

Quantization made by Richard Erkhov.

Github

Discord

Request more models

phi-2-super - GGUF

Model creator: https://huggingface.co/abacaj/
Original model: https://huggingface.co/abacaj/phi-2-super/

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

Original model description:

license: mit license_link: https://huggingface.co/microsoft/phi-2/resolve/main/LICENSE language:

en widget:
text: Hello who are you? example_title: Identity
text: What can you do? example_title: Capabilities
text: Create a fastapi endpoint to retrieve the weather given a zip code. example_title: Coding tags:
convAI
conversational pipeline_tag: text-generation model-index:
name: phi-2-super results:
IFEval
- task: type: text-generation name: Text Generation dataset: name: Instruction Following Eval type: wis-k/instruction-following-eval metrics:
  - type: acc name: prompt_level_loose_acc value: 0.2717 source: name: LightEval url: https://github.com/huggingface/lighteval

Phi-2-super (SFT + cDPO)

Base Model: microsoft/phi-2

How to run inference:

python

import transformers
import torch

if __name__ == "__main__":
  model_name = "abacaj/phi-2-super"
  tokenizer = transformers.AutoTokenizer.from_pretrained(model_name)
  
  model = (
      transformers.AutoModelForCausalLM.from_pretrained(
          model_name,
      )
      .to("cuda:0")
      .eval()
  )
  
  messages = [
      {"role": "user", "content": "Hello, who are you?"}
  ]
  inputs = tokenizer.apply_chat_template(messages, return_tensors="pt").to(model.device)
  input_ids_cutoff = inputs.size(dim=1)
  
  with torch.no_grad():
      generated_ids = model.generate(
          input_ids=inputs,
          use_cache=True,
          max_new_tokens=512,
          temperature=0.2,
          top_p=0.95,
          do_sample=True,
          eos_token_id=tokenizer.eos_token_id,
          pad_token_id=tokenizer.pad_token_id,
      )
  
  completion = tokenizer.decode(
      generated_ids[0][input_ids_cutoff:],
      skip_special_tokens=True,
  )
  
  print(completion)

Chat template

The model uses the same chat template as found in Mistral instruct models:

python

text = "<|endoftext|>[INST] What is your favourite condiment? [/INST]"
"Well, I'm quite partial to a good squeeze of fresh lemon juice. It adds just the right amount of zesty flavour to whatever I'm cooking up in the kitchen!<|endoftext|> "
"[INST] Do you have mayonnaise recipes? [/INST]"

You don't need to do it manually if you use the HF transformers tokenizer:

python

  messages = [
      {"role": "user", "content": "Hello, who are you?"},
      {"role": "assistant": "content": "I am ..."}
  ]
  inputs = tokenizer.apply_chat_template(messages, return_tensors="pt").to(model.device)

RichardErkhov/abacaj_-_phi-2-super-gguf

Original model description:

IFEval

Phi-2-super (SFT + cDPO)

How to run inference:

Chat template

MT-bench / heval

RichardErkhov/
abacaj_-_phi-2-super-gguf