Code

Llama 3.2-Vision Implementation#1160

Code

Open

mktexan wants to merge 9 commits intoSkyvern-AI:mainSkyvern-AI/skyvern:mainfrom mktexan:Llamamktexan/skyvern:LlamaCopy head branch name to clipboard

+1,163-175

Commits

Commits on Nov 9, 2024

• Adding in support for Llama 3.2

  mktexan

  committed
  36403e3

  Copy full SHA for 36403e3

• Data return instruction updates for Ollama. Added function for markdown format fix for Ollama.

  mktexan

  committed
  aecaa23

  Copy full SHA for aecaa23

• Provided more explicit instructions to Llama on its expected output.

  mktexan

  committed
  6c646b5

  Copy full SHA for 6c646b5

• Testing llama specific instructions.

  mktexan

  committed
  35d94e0

  Copy full SHA for 35d94e0

• Refactor of promtps. Temporary disabling of skyvern prompts for Llama testing.

  mktexan

  committed
  44a04cd

  Copy full SHA for 44a04cd

Commits on Nov 10, 2024

• Further testing of custom prompts for llama.

  mktexan

  committed
  4219ecf

  Copy full SHA for 4219ecf

• Switched LLM server. Fine tuned extract-actions for llama.

  mktexan

  committed
  4b3ee6b

  Copy full SHA for 4b3ee6b

Commits on Nov 11, 2024

• Llama Extract-action outputting json response.

  mktexan

  committed
  d14dd20

  Copy full SHA for d14dd20

Commits on Nov 12, 2024

• Added utils function to compensate for LLama conversational responses. Closer to success with extract-action.

  mktexan

  committed
  ba837f3

  Copy full SHA for ba837f3