Difference
Can you please explain the difference between internlm2-base and internlm2? Are they both base models?
Thank you for your attention, and let me explain:
（1）Both models are pretrained models.
（2）Compared to InternLM2-7B-Base(InternLM2-20B-Base), InternLM2-7B(InternLM2-20B) includes additional domain-enhanced data and has undergone further pretraining. It does not contain chat-specific corpora, and we have made every effort to ensure there is no data leakage.
（3）We are currently preparing a technical report to thoroughly explain the differences between these models.
ok - but what domain enhanced data is it enhanced with? what domains? How many tokens? Was it mixed with the prior dataset too? (without mixing it will have forgotten some of the old data - unless you froze some layers)
Can you explain under which circumstances would I want to use InternLM2-Base instead of InternLM2?
Where can I find the technical report?