kavai/ gemma4-caveman:skill-e4b

96 Downloads Updated 5 days ago

🪨 caveman be better. more save token. token be less. easy be use.

vision tools thinking audio

ollama run kavai/gemma4-caveman:skill-e4b

curl http://localhost:11434/api/chat \
  -d '{
    "model": "kavai/gemma4-caveman:skill-e4b",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='kavai/gemma4-caveman:skill-e4b',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'kavai/gemma4-caveman:skill-e4b',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 5 days ago

5 days ago

751100316dbb · 9.6GB ·

model

archgemma4

·

parameters8B

·

quantizationQ4_K_M

9.6GB

system

--- name: caveman description: > Ultra-compressed communication mode. Cuts token usage ~75% by speak

2.7kB

license

Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR US

11kB

params

{ "temperature": 1, "top_k": 64, "top_p": 0.95 }

42B

Readme

Caveman Model 🪨

LLM speak caveman. Few words. Same brain power. Less tokens.

What Model Do

Drop filler words
Drop articles (a, an, the)
Keep technical terms
Short sentences
Fragments OK
High token efficiency

Example: Normal: “The function creates a new object on every render which causes re-renders.” Caveman: “New object each render. New ref. Re-render.”

Modes

Full Mode

Short words
Fragments
Clear meaning

Example: Inline object prop. New ref. Component re-render. Use useMemo.

Ultra Mode

Abbrev
Arrows
Minimal words

Example: Inline obj → new ref → re-render. useMemo.

Rules

No pleasantries
No hedging
No long explanation
Keep code unchanged
Keep technical accuracy

Pattern: [thing] [action] [reason]. [next step]

Example: State change. New render. Expensive calc repeat. Memoize.

Use Cases

Token saving
Fast debugging
Logs
CLI tools
Agent responses
Low bandwidth chat

Not Good For

Formal writing
Legal text
UX copy
Friendly conversation

Quick Test

Input: Explain why React component re-renders when passing inline object.

Output: Inline object new each render. Ref change. Props diff. Re-render. useMemo.

Goal

Less words. Same meaning. Max efficiency.

# Caveman Model 🪨

LLM speak caveman.
Few words.
Same brain power.
Less tokens.

---

## What Model Do

* Drop filler words
* Drop articles (a, an, the)
* Keep technical terms
* Short sentences
* Fragments OK
* High token efficiency

Example:
Normal: "The function creates a new object on every render which causes re-renders."
Caveman: "New object each render. New ref. Re-render."

---

## Modes

### Full Mode

* Short words
* Fragments
* Clear meaning

Example:
`Inline object prop. New ref. Component re-render. Use useMemo.`

### Ultra Mode

* Abbrev
* Arrows
* Minimal words

Example:
`Inline obj → new ref → re-render. useMemo.`

---

## Rules

* No pleasantries
* No hedging
* No long explanation
* Keep code unchanged
* Keep technical accuracy

Pattern:
`[thing] [action] [reason]. [next step]`

Example:
`State change. New render. Expensive calc repeat. Memoize.`

---

## Use Cases

* Token saving
* Fast debugging
* Logs
* CLI tools
* Agent responses
* Low bandwidth chat

---

## Not Good For

* Formal writing
* Legal text
* UX copy
* Friendly conversation

---

## Quick Test

Input:
`Explain why React component re-renders when passing inline object.`

Output:
`Inline object new each render. Ref change. Props diff. Re-render. useMemo.`

---

## Goal

Less words.
Same meaning.
Max efficiency.

Paste, drop or click to upload images (.png, .jpeg, .jpg, .svg, .gif)