Here is my current situation. I created another branch and modified my code on the branch. Then I git commmit on the branch.
Well, yes and no. See What exactly do we mean by "branch"?
As in the comments below your question, you did:
git checkout <some-hash-ID>
which checks out a specific commit, as a sort of go back in time to when this was the current commit operation. Git calls this kind of git checkout a "detached HEAD", and prints a huge warning (in any modern Git—very old Git, 1.7 era, was pretty quiet about it, as I recall), because it's easy to start committing away in this situation.
You then made new commits. In Git, this is a thing you can do, even if you're not on any branch, i.e., if you're in this detached HEAD mode. Git goes ahead and makes new commits, and—as in that other question—those new commits form one kind of branch, what I like to sometimes call a DAGlet.
Since there's no name involved, though, the only name you have for these new commits is their raw hash IDs, or—at least for the moment—HEAD. The name HEAD works up until you re-attach your HEAD by running, e.g., git checkout master.
Once you've reattached your HEAD (so that HEAD just means master which then means whichever commit is the tip of the branch named master), the name HEAD stops working to mean the commits you made while you were carrying your HEAD around in your arms, like some sort of ghoul chasing Ichabod Crane.
We can draw the situation like this. You probably did git checkout 7830210, so we know that commit's hash ID; I think e4c58... is the new commit you created. There are more commits whose ID we don't know, and really don't care about, which we can represent as round dots, o. So while your HEAD was rolling around loosely :-) you had this:
...--o--7830210...--o--o <-- master
\
e4c58... <-- HEAD
You then told Git to re-attach HEAD to master, giving:
...--o--7830210...--o--o <-- master (HEAD)
\
e4c58... [no name -- abandoned]
To manipulate the new commit more generally, you will need to give it a name. Git being Git, there are many ways to do that. The simplest at this point is to use git branch, which can create a new branch when given a commit hash ID:
$ git branch xyzzy-branch e4c58
which gives us:
...--o--7830210...--o--o <-- master (HEAD)
\
e4c58... <-- xyzzy-branch
(note that your HEAD remains untouched here, so it's still attached to master).
DAGlet is an invented word, deliberately punning from both the real word aglet, which means one of those plastic-covered shoelace tips, for instance, and DAG plus the diminutive suffix -let. DAG is a common computer science acronym for Directed Acyclic Graph. A DAGlet is just any useful snippet of a bigger DAG. We only care about snippets that end with a name pointing to it, or—in some cases—with no name pointing to them any more, as in this particular case. By wrapping them up in metaphorical plastic, we make it easy to grab hold of them and use them: voila, DAGlets.)
As far as I know, I invented this term. I see the word DAGlet also used in an IEEE paper, but I used it earlier. (It's also part of the name of a place—it appears to be a small oasis in the Sahara—in Algeria: Daglet Lagrouba.) Still, the diminutive suffix is kind of an obvious formation.
