Datacenter Models
Collection
Lmao • 2 items • Updated
# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("ConicCat/GLM-4.7-Architect-355B-A32B")
model = AutoModelForCausalLM.from_pretrained("ConicCat/GLM-4.7-Architect-355B-A32B", device_map="auto")
messages = [
{"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
messages,
add_generation_prompt=True,
tokenize=True,
return_dict=True,
return_tensors="pt",
).to(model.device)
outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))Quick Links
ConicCat/GLM-4.7-Architect-355B-A32B
A finetune of GLM-4.5 air to improve prose and writitquality and attempt to remove the bulk of glm-isms using a Gutenberg-like methodology.
No particular attempt was made to preverse thinking ability; I recommend skipping thinking as in the GLM template i.e. using </think> as a prefill.
Trained on a variety of backtranslated critically acclaimed short story anthologies for 8 epochs.
- Downloads last month
- 24
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="ConicCat/GLM-4.7-Architect-355B-A32B") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)