RWKV-LM
GitHubRWKV (pronounced RwaKuv) is an RNN with great LLM performance, which can also be directly trained like a GPT transformer (parallelizable). We are at RWKV-7 "Goose". So it's combining the best of RNN and transformer - great performance, linear time, constant space (no kv-cache), fast training, infinite ctx_len, and free sentence embedding.
Claude Desktop Configuration
Add to ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"RWKV-LM": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-RWKV-LM"
]
}
}
}