RichardErkhov/cloudyu_-_Yi-34Bx2-MOE-200K-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Yi-34Bx2-MOE-200K - GGUF

• Model creator: https://huggingface.co/cloudyu/
• Original model: https://huggingface.co/cloudyu/Yi-34Bx2-MOE-200K/

Name	Quant method	Size
Yi-34Bx2-MOE-200K.Q2_K.gguf	Q2_K	20.86GB
Yi-34Bx2-MOE-200K.IQ3_XS.gguf	IQ3_XS	23.26GB
Yi-34Bx2-MOE-200K.IQ3_S.gguf	IQ3_S	24.56GB
Yi-34Bx2-MOE-200K.Q3_K_S.gguf	Q3_K_S	24.51GB
Yi-34Bx2-MOE-200K.IQ3_M.gguf	IQ3_M	25.2GB
Yi-34Bx2-MOE-200K.Q3_K.gguf	Q3_K	27.23GB
Yi-34Bx2-MOE-200K.Q3_K_M.gguf	Q3_K_M	27.23GB
Yi-34Bx2-MOE-200K.Q3_K_L.gguf	Q3_K_L	29.59GB
Yi-34Bx2-MOE-200K.IQ4_XS.gguf	IQ4_XS	30.58GB
Yi-34Bx2-MOE-200K.Q4_0.gguf	Q4_0	31.98GB
Yi-34Bx2-MOE-200K.IQ4_NL.gguf	IQ4_NL	32.27GB
Yi-34Bx2-MOE-200K.Q4_K_S.gguf	Q4_K_S	32.22GB
Yi-34Bx2-MOE-200K.Q4_K.gguf	Q4_K	34.14GB
Yi-34Bx2-MOE-200K.Q4_K_M.gguf	Q4_K_M	34.14GB
Yi-34Bx2-MOE-200K.Q4_1.gguf	Q4_1	35.49GB
Yi-34Bx2-MOE-200K.Q5_0.gguf	Q5_0	39.0GB
Yi-34Bx2-MOE-200K.Q5_K_S.gguf	Q5_K_S	39.0GB
Yi-34Bx2-MOE-200K.Q5_K.gguf	Q5_K	40.12GB
Yi-34Bx2-MOE-200K.Q5_K_M.gguf	Q5_K_M	40.12GB
Yi-34Bx2-MOE-200K.Q5_1.gguf	Q5_1	42.51GB
Yi-34Bx2-MOE-200K.Q6_K.gguf	Q6_K	46.47GB
Yi-34Bx2-MOE-200K.Q8_0.gguf	Q8_0	60.18GB

Original model description:

license: other

---

try to build a 200K context length MoE Yi based chat model.

[[GGUF 4bit is here ]https://huggingface.co/cloudyu/Yi-34Bx2-MOE-200K-gguf ]

• Metrics

metrics :

code example

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
import math

## 
model_path = "cloudyu/Yi-34Bx2-MOE-200K"

tokenizer = AutoTokenizer.from_pretrained(model_path, use_default_system_prompt=False)
model = AutoModelForCausalLM.from_pretrained(
    model_path, torch_dtype=torch.float32, device_map='auto',local_files_only=False, load_in_4bit=True
)
print(model)
prompt = input("please input prompt:")
while len(prompt) > 0:
  input_ids = tokenizer(prompt, return_tensors="pt").input_ids.to("cuda")

  generation_output = model.generate(
    input_ids=input_ids, max_new_tokens=500,repetition_penalty=1.2
  )
  print(tokenizer.decode(generation_output[0]))
  prompt = input("please input prompt:")