Reinforcement Learning
Transformers
Safetensors
olmo2
text-generation
rlvr
rl-zero
chain-of-thought
faithfulness
reward-hacking
cue-injection
unfaithrl
Instructions to use UnfaithRL/OLMo-2-0425-1B-hybrid_accuracy_hint_reward-2048 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use UnfaithRL/OLMo-2-0425-1B-hybrid_accuracy_hint_reward-2048 with Transformers:
# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("UnfaithRL/OLMo-2-0425-1B-hybrid_accuracy_hint_reward-2048") model = AutoModelForCausalLM.from_pretrained("UnfaithRL/OLMo-2-0425-1B-hybrid_accuracy_hint_reward-2048", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download chat_template.jinja from UnfaithRL/OLMo-2-0425-1B-hybrid_accuracy_hint_reward-2048: direct link, hf CLI and curl.
- Browser
- Download file 508 Bytes
-
https://huggingface.co/UnfaithRL/OLMo-2-0425-1B-hybrid_accuracy_hint_reward-2048/resolve/main/chat_template.jinja
- Command line
-
hf download hf://UnfaithRL/OLMo-2-0425-1B-hybrid_accuracy_hint_reward-2048/chat_template.jinja
-
curl -L -o chat_template.jinja https://huggingface.co/UnfaithRL/OLMo-2-0425-1B-hybrid_accuracy_hint_reward-2048/resolve/main/chat_template.jinja
508 Bytes
| {{ bos_token }}{% for message in messages %}{% if message['role'] == 'system' %}{{ '<|system|> | |
| ' + message['content'] + ' | |
| ' }}{% elif message['role'] == 'user' %}{{ '<|user|> | |
| ' + message['content'] + ' | |
| ' }}{% elif message['role'] == 'assistant' %}{% if not loop.last %}{{ '<|assistant|> | |
| ' + message['content'] + eos_token + ' | |
| ' }}{% else %}{{ '<|assistant|> | |
| ' + message['content'] + eos_token }}{% endif %}{% endif %}{% if loop.last and add_generation_prompt %}{{ '<|assistant|> | |
| ' }}{% endif %}{% endfor %} |