v0.12.0

• [88d882c] build(make): bump version to 0.12.0 (#52)
• [b417b52] build(docker): bump version to 0.12.0 (#51)
• [6d03b98] build(docker): bundle a patched version of chatglm-6b (#55)
• [0f3c44c] feat(model): allow to load a specific revision (#54)
• [50521fd] build(utils): allow to download a specific revision (#53)
• [a0bc6c3] fix(model): override torch dtype in 8-bit mode (#50)
• [056d25a] build(docker): add example for bundling chatglm-6b (#49)
• [9933b77] build(deps): update transformers[sentencepiece] requirement (#46)
• [d10eb70] feat(model): add support for half-precision (#45)