cactus-compute/needle

• highreadme#1

  Reposition README H1 to explicitly state it's an LLM

  原因：

  当前

  # Needle

  复制粘贴的修复

  复制

  # Needle: A 26M Parameter On-Device LLM for Function Calling

  待办完成忽略
• mediumreadme#2

  Add a concise 'About' section to the README

  原因：

  复制粘贴的修复

  复制

  ## About Needle
  
  Needle is a highly efficient, 26 million parameter (26M) function call model, distilled from Gemini 3.1. Designed for on-device AI, it runs at high speeds on incredibly small devices and can be finetuned locally on personal computers. It's ideal for applications requiring compact, performant language models.

  待办完成忽略
• lowreadme#3

  Add a 'Key Features' section emphasizing on-device capabilities

  原因：

  复制粘贴的修复

  复制

  ## Key Features
  
  *   **Ultra-Compact:** A 26M parameter model, making it one of the smallest function call LLMs available.
  *   **On-Device Performance:** Achieves 6000 toks/sec prefill and 1200 decode speed on small hardware.
  *   **Local Finetuning:** Easily finetune the model on your personal Mac/PC.
  *   **Open Weights & Dataset:** Fully open-source weights and dataset generation for transparency and community contribution.

  待办完成忽略

本次扫描解析到的品类 GEO 通道：google/gemini-2.5-flash, deepseek/deepseek-v4-flash

• 品类问题

  What are efficient, open-source language models for on-device inference on small hardware?

  你：未被推荐
  AI 推荐顺序：

  1. TinyLlama 1.1B
  2. Phi-2
  3. Mistral 7B Instruct v0.2
  4. Gemma 2B Instruct
  5. OpenLLaMA 3B/7B
  6. Pythia 160M/410M/1B

  AI 推荐了 6 个替代方案，却始终没点名 cactus-compute/needle。这就是要补上的差距。

  查看 AI 完整回答
• 品类问题

  Seeking a compact LLM that can be finetuned locally on a personal computer.

  你：未被推荐
  AI 推荐顺序：

  1. Mistral 7B
  2. Llama 2 7B
  3. TinyLlama 1.1B
  4. Phi-2
  5. Gemma 2B
  6. Gemma 7B
  7. Falcon 7B

  AI 推荐了 7 个替代方案，却始终没点名 cactus-compute/needle。这就是要补上的差距。

  查看 AI 完整回答

• Metadata completeness

  pass

• README presence

  pass

• Compared to common alternatives in this category, what is the core differentiator of cactus-compute/needle?

  pass
  ✓AI 明确点名了 cactus-compute/needle

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？

• If a team adopts cactus-compute/needle in production, what risks or prerequisites should they evaluate first?

  pass
  ✓AI 明确点名了 cactus-compute/needle

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？

• In one sentence, what problem does the repo cactus-compute/needle solve, and who is the primary audience?

  pass
  ✓AI 明确点名了 cactus-compute/needle

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？