emre şahin's digital garden 🍃 - LLM https://emresahin.net/tags/llm/ Posts in the LLM tag en contact@emresahin.net (Emre Şahin) Tue, 15 Sep 2026 19:46:32 +0000 bits 21 2025-09-28T18:00:14+00:00 2025-09-28T18:00:14+00:00 Emre Şahin Sun, 28 Sep 2025 18:00:14 +0000 https://emresahin.net/bits-21/ https://emresahin.net/bits-21/ I wanted to get the number of word changes between commits in my blog. I asked for a git-diff-based Nushell pipeline from Gemini. Gemini proposed this: git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length What I ended up ... bits programming Gemini nushell git regex LLM I wanted to get the number of word changes between commits in my blog. I asked for a git-diff-based Nushell pipeline from Gemini.

Gemini proposed this:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length

What I ended up doing:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -a "{+" "" | str replace -a "+}" "" | str replace -a ' ' "\n" | lines | length

I notice LLMs tend to err on the side of complexity. I don’t know if this is due to their training to spit out as many tokens as possible, but using regular expressions when a simple string replacement may suffice is a good sign that they may be adding more complexity than necessary.

]]> bits 20 2025-09-28T17:52:31+00:00 2025-09-28T17:52:31+00:00 Emre Şahin Sun, 28 Sep 2025 17:52:31 +0000 https://emresahin.net/bits-20/ https://emresahin.net/bits-20/ I wanted to get the number of word changes between commits in my blog. I asked for a git-diff-based Nushell pipeline from Gemini. Gemini proposed this: git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length What I ended up ... bits programming Gemini nushell git regex LLM I wanted to get the number of word changes between commits in my blog. I asked for a git-diff-based Nushell pipeline from Gemini.

Gemini proposed this:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length

What I ended up doing:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -a "{+" "" | str replace -a "+}" "" | str replace -a ' ' "\n" | lines | length

I notice LLMs tend to err on the side of complexity. I don’t know if this is due to their training to spit out as many tokens as possible, but using regular expressions when a simple string replacement may suffice is a good sign that they may be adding more complexity than necessary.

]]> bits 19 2025-09-28T17:42:11+00:00 2025-09-28T17:42:11+00:00 Emre Şahin Sun, 28 Sep 2025 17:42:11 +0000 https://emresahin.net/bits-19/ https://emresahin.net/bits-19/ I wanted to get the number of word changes between commits in my blog. I asked for a git-diff-based Nushell pipeline from Gemini. Gemini proposed this: git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length What I ended up ... bits programming Gemini nushell git regex LLM I wanted to get the number of word changes between commits in my blog. I asked for a git-diff-based Nushell pipeline from Gemini.

Gemini proposed this:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length

What I ended up doing:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -a "{+" "" | str replace -a "+}" "" | str replace -a ' ' "\n" | lines | length

I notice LLMs tend to err on the side of complexity. I don’t know if this is due to their training to spit out as many tokens as possible, but using regular expressions when a simple string replacement may suffice is a good sign that they may be adding more complexity than necessary.

]]> bits 18 2025-09-28T17:22:48+00:00 2025-09-28T17:22:48+00:00 Emre Şahin Sun, 28 Sep 2025 17:22:48 +0000 https://emresahin.net/bits-18/ https://emresahin.net/bits-18/ I wanted to get the number of word changes between commits in my blog. I asked for a git-diff-based Nushell pipeline from Gemini. Gemini proposed this: git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length What I ended up ... bits programming Gemini nushell git regex LLM I wanted to get the number of word changes between commits in my blog. I asked for a git-diff-based Nushell pipeline from Gemini.

Gemini proposed this:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length

What I ended up doing:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -a "{+" "" | str replace -a "+}" "" | str replace -a ' ' "\n" | lines | length

I notice LLMs tend to err on the side of complexity. I don’t know if this is due to their training to spit out as many tokens as possible, but using regular expressions when a simple string replacement may suffice is a good sign that they may be adding more complexity than necessary.

]]> bits 16 2025-09-28T17:15:52+00:00 2025-09-28T17:15:52+00:00 Emre Şahin Sun, 28 Sep 2025 17:15:52 +0000 https://emresahin.net/bits-16/ https://emresahin.net/bits-16/ I wanted to get the number of word changes between commits in my blogs. I asked for a git diff based nushell pipeline from Gemini: Gemini proposed this: git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length What I ended up... bits Gemini nushell git regex LLM workflow I wanted to get the number of word changes between commits in my blogs. I asked for a git diff based nushell pipeline from Gemini:

Gemini proposed this:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -r '\{\+|\+\}' '' | str join " " | str words | length

What I ended up doing:

git diff --word-diff=plain -- "*.md" | rg -o '\{\+(.*?)\+\}' | str replace -a "{+" "" | str replace -a "+}" "" | str replace -a ' ' "\n" | lines | length

I notice LLMs tend to err on the side of complexity. I don’t know if this is due to their training to spit as many tokens as possible, but using regular expressions when a simple single replace may suffice is a good sign that they may be adding more complexity than necessary.

]]>