0. Backup to volume
    a. Volume can be found in `/vol_c`
    b. directories to back up
         - `raw_data`, `data`, `mfcc`, `exp`
    c. command to use
         - `cp -r [dir_to_copy] [path_to_volume]`
1. General Questions/Comments
2. Big picture contribution of LM
3. Brief summary of language modeling process: from corpus to LM
4. Smoothing
    a. Why we need it
    b. Which approach to use
5. IRSTLM
    a. manual: http://hermes.fbk.eu/people/bertoldi/te … manual.pdf
    b. already compiled and code in `/scratch/kaldi/tools/irstlm/bin` (must view from *inside* `docker`!)
    c. how to use (see notebook 2.1)
6. Next week's HW
    a. submission
        - `File -> Download As -> HTML`
    b. copy template
7. Generating the probability of a sequence (see notebook 2.2)
    a. Default situation (len(sequence) <= size(n-gram) and n-gram in LM)
    b. Two special situations
        - n-gram not in LM
        - sequence is larger than n-gram
8. "ate the lion" v. "ate the mouse" problem
    a. Why did it happen?     
9. Impact of LM n-gram size
10. Impact of LM size
    a. space
    b. speed
    c. alternative options
        - pruning (IRSTLM manual, section 5)
           - hyperparameter = ???
        - rescoring
11. Intuitions about ARPA-style LMs
12. n-gram LM v. RNN
13.  What to expect next week
    a. Week 3 items
        1. `kaldi_config.json` usage
        2. Building the `data` directory
    b. Week 2 HW
        1. identifying a case study
        2. reviewing others' case studies