Open nfwyst opened 3 months ago
config:
return { "huggingface/llm.nvim", opts = { model = "rouge/autocoder-s-6.7b:latest", backend = "ollama", url = "http://localhost:11434/api/generate", request_body = { parameters = { max_new_tokens = 100000, -- the maximum numbers of tokens to generate, ignore the number of prompt token temperature = 0.2, -- the bigger, more creatively top_p = 0.95, -- the bigger, text generated is diverse }, }, tokens_to_clear = { "<|endoftext|>" }, fim = { enabled = true, prefix = "<fim_prefix>", middle = "<fim_middle>", suffix = "<fim_suffix>", }, accept_keymap = "<Tab>", dismiss_keymap = "<S-Tab>", lsp = { bin_path = vim.api.nvim_call_function("stdpath", { "data" }) .. "/mason/bin/llm-ls", }, context_window = 100000, -- max number of tokens for the context window tokenizer = { repository = "Bin12345/AutoCoder_S_6.7B", }, }, }
error message: [LLM] serde json error: EOF while parsing a value at line 1 column 0
Your request body is incorrect, refer to https://github.com/huggingface/llm.nvim?tab=readme-ov-file#ollama and https://github.com/ollama/ollama/blob/main/docs/api.md#request-6 for more info.
config:
error message: [LLM] serde json error: EOF while parsing a value at line 1 column 0