Instructions to use AICube/ChatGLM with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use AICube/ChatGLM with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("THUDM/chatglm3-6b-base") model = PeftModel.from_pretrained(base_model, "AICube/ChatGLM") - Notebooks
- Google Colab
- Kaggle
File size: 929 Bytes
5083c9c | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 | {
"added_tokens_decoder": {},
"auto_map": {
"AutoTokenizer": [
"tokenization_chatglm.ChatGLMTokenizer",
null
]
},
"chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ system_message + '\\n' }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ 'Human: ' + content + '\\nAssistant: ' }}{% elif message['role'] == 'assistant' %}{{ content + '</s>' + '\\n' }}{% endif %}{% endfor %}",
"clean_up_tokenization_spaces": false,
"do_lower_case": false,
"encode_special_tokens": false,
"eos_token": "</s>",
"model_max_length": 1000000000000000019884624838656,
"pad_token": "<unk>",
"padding_side": "right",
"remove_space": false,
"split_special_tokens": false,
"tokenizer_class": "ChatGLMTokenizer",
"unk_token": "<unk>"
}
|