name: Qwen-2.5-base-tron-7b
merge_method: sce
parameters:
  select_topk: 0.666
  normalize: true
dtype: float32
out_dtype: bfloat16
base_model: Jebadiah/Qwen-2.5-base-7b
tokenizer:
  source: union
  special_tokens: keep_all
  priority: none
  add_padding_token: true
  force_fast_tokenizer: true  # Can help with compatibility
  resolve_conflicts: append_ids  # Append IDs to conflicting tokens to make them unique
models:
  - model: bunnycore/Blabbertron-1.2
  - model: Xiaojian9992024/Qwen2.5-7B-MS-Destroyer
  - model: trollek/Qwen2.5-7B-CySecButler-v0.1

💻 Usage

python

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Jebadiah/Qwen-2.5-base-tron-7b"
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"])