{ "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import sys\n", "sys.path.append(\"..\")\n", "import config" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "PROMPT = \"\"\"你是一位精通简体中文的专业翻译，尤其擅长将专业学术论文翻译成浅显易懂的科普文章。请你帮我将以下英文段落翻译成中文，风格与中文科普读物相似。\n", "\n", "规则：\n", "- 翻译时要准确传达原文的事实和背景。\n", "- 即使上意译也要保留原始段落格式，以及保留术语，例如 FLAC，JPEG 等。保留公司缩写，例如 Microsoft, Amazon, OpenAI 等。\n", "- 人名不翻译\n", "- 全角括号换成半角括号，并在左括号前面加半角空格，右括号后面加半角空格。\n", "- 在翻译专业术语时，第一次出现时要在括号里面写上英文原文，例如：“生成式 AI (Generative AI)”，之后就可以只写中文了。\n", "- 以下是常见的 AI 相关术语词汇对应表（English -> 中文）：\n", " * Transformer -> Transformer\n", " * Token -> Token\n", " * LLM/Large Language Model -> 大语言模型\n", " * Zero-shot -> 零样本\n", " * Few-shot -> 少样本\n", " * AI Agent -> AI 智能体\n", " * AGI -> 通用人工智能\"\"\"" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [], "source": [ "from transcribe.helpers.translator import QwenTranslator" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "llama_context: n_ctx_per_seq (512) < n_ctx_train (131072) -- the full capacity of the model will not be utilized\n", "ggml_metal_init: skipping kernel_get_rows_bf16 (not supported)\n", "ggml_metal_init: skipping kernel_mul_mv_bf16_f32 (not supported)\n", "ggml_metal_init: skipping kernel_mul_mv_bf16_f32_1row (not supported)\n", "ggml_metal_init: skipping kernel_mul_mv_bf16_f32_l4 (not supported)\n", "ggml_metal_init: skipping kernel_mul_mv_bf16_bf16 (not supported)\n", "ggml_metal_init: skipping kernel_mul_mv_id_bf16_f32 (not supported)\n", "ggml_metal_init: skipping kernel_mul_mm_bf16_f32 (not supported)\n", "ggml_metal_init: skipping kernel_mul_mm_id_bf16_f32 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h64 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h80 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h96 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h112 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h128 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h192 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_hk192_hv128 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h256 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_h96 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_h128 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_h192 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_hk192_hv128 (not supported)\n", "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_h256 (not supported)\n", "ggml_metal_init: skipping kernel_cpy_f32_bf16 (not supported)\n", "ggml_metal_init: skipping kernel_cpy_bf16_f32 (not supported)\n", "ggml_metal_init: skipping kernel_cpy_bf16_bf16 (not supported)\n" ] } ], "source": [ "translator = QwenTranslator(config.LLM_LARGE_MODEL_PATH, PROMPT, PROMPT)" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "ename": "TypeError", "evalue": "'<=' not supported between instances of 'str' and 'int'", "output_type": "error", "traceback": [ "\u001b[31m---------------------------------------------------------------------------\u001b[39m", "\u001b[31mTypeError\u001b[39m Traceback (most recent call last)", "\u001b[36mCell\u001b[39m\u001b[36m \u001b[39m\u001b[32mIn[7]\u001b[39m\u001b[32m, line 1\u001b[39m\n\u001b[32m----> \u001b[39m\u001b[32m1\u001b[39m \u001b[43mtranslator\u001b[49m\u001b[43m(\u001b[49m\u001b[33;43m\"\u001b[39;49m\u001b[33;43m欢迎来到go sim大会\u001b[39;49m\u001b[33;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[33;43m\"\u001b[39;49m\u001b[33;43mzh\u001b[39;49m\u001b[33;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[33;43m'\u001b[39;49m\u001b[33;43men\u001b[39;49m\u001b[33;43m'\u001b[39;49m\u001b[43m)\u001b[49m\n", "\u001b[36mFile \u001b[39m\u001b[32m~/Codes/trans/tests/../transcribe/helpers/translator.py:30\u001b[39m, in \u001b[36mQwenTranslator.__call__\u001b[39m\u001b[34m(self, prompt, *args, **kwargs)\u001b[39m\n\u001b[32m 29\u001b[39m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34m__call__\u001b[39m(\u001b[38;5;28mself\u001b[39m, prompt,*args, **kwargs):\n\u001b[32m---> \u001b[39m\u001b[32m30\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mllm\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m 31\u001b[39m \u001b[43m \u001b[49m\u001b[43mprompt\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 32\u001b[39m \u001b[43m \u001b[49m\u001b[43m*\u001b[49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 33\u001b[39m \u001b[43m \u001b[49m\u001b[43m*\u001b[49m\u001b[43m*\u001b[49m\u001b[43mkwargs\u001b[49m\n\u001b[32m 34\u001b[39m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n", "\u001b[36mFile \u001b[39m\u001b[32m~/Codes/trans/.venv/lib/python3.11/site-packages/llama_cpp/llama.py:1902\u001b[39m, in \u001b[36mLlama.__call__\u001b[39m\u001b[34m(self, prompt, suffix, max_tokens, temperature, top_p, min_p, typical_p, logprobs, echo, stop, frequency_penalty, presence_penalty, repeat_penalty, top_k, stream, seed, tfs_z, mirostat_mode, mirostat_tau, mirostat_eta, model, stopping_criteria, logits_processor, grammar, logit_bias)\u001b[39m\n\u001b[32m 1838\u001b[39m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34m__call__\u001b[39m(\n\u001b[32m 1839\u001b[39m \u001b[38;5;28mself\u001b[39m,\n\u001b[32m 1840\u001b[39m prompt: \u001b[38;5;28mstr\u001b[39m,\n\u001b[32m (...)\u001b[39m\u001b[32m 1864\u001b[39m logit_bias: Optional[Dict[\u001b[38;5;28mint\u001b[39m, \u001b[38;5;28mfloat\u001b[39m]] = \u001b[38;5;28;01mNone\u001b[39;00m,\n\u001b[32m 1865\u001b[39m ) -> Union[CreateCompletionResponse, Iterator[CreateCompletionStreamResponse]]:\n\u001b[32m 1866\u001b[39m \u001b[38;5;250m \u001b[39m\u001b[33;03m\"\"\"Generate text from a prompt.\u001b[39;00m\n\u001b[32m 1867\u001b[39m \n\u001b[32m 1868\u001b[39m \u001b[33;03m Args:\u001b[39;00m\n\u001b[32m (...)\u001b[39m\u001b[32m 1900\u001b[39m \u001b[33;03m Response object containing the generated text.\u001b[39;00m\n\u001b[32m 1901\u001b[39m \u001b[33;03m \"\"\"\u001b[39;00m\n\u001b[32m-> \u001b[39m\u001b[32m1902\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mcreate_completion\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m 1903\u001b[39m \u001b[43m \u001b[49m\u001b[43mprompt\u001b[49m\u001b[43m=\u001b[49m\u001b[43mprompt\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1904\u001b[39m \u001b[43m \u001b[49m\u001b[43msuffix\u001b[49m\u001b[43m=\u001b[49m\u001b[43msuffix\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1905\u001b[39m \u001b[43m \u001b[49m\u001b[43mmax_tokens\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmax_tokens\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1906\u001b[39m \u001b[43m \u001b[49m\u001b[43mtemperature\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtemperature\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1907\u001b[39m \u001b[43m \u001b[49m\u001b[43mtop_p\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtop_p\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1908\u001b[39m \u001b[43m \u001b[49m\u001b[43mmin_p\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmin_p\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1909\u001b[39m \u001b[43m \u001b[49m\u001b[43mtypical_p\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtypical_p\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1910\u001b[39m \u001b[43m \u001b[49m\u001b[43mlogprobs\u001b[49m\u001b[43m=\u001b[49m\u001b[43mlogprobs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1911\u001b[39m \u001b[43m \u001b[49m\u001b[43mecho\u001b[49m\u001b[43m=\u001b[49m\u001b[43mecho\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1912\u001b[39m \u001b[43m \u001b[49m\u001b[43mstop\u001b[49m\u001b[43m=\u001b[49m\u001b[43mstop\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1913\u001b[39m \u001b[43m \u001b[49m\u001b[43mfrequency_penalty\u001b[49m\u001b[43m=\u001b[49m\u001b[43mfrequency_penalty\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1914\u001b[39m \u001b[43m \u001b[49m\u001b[43mpresence_penalty\u001b[49m\u001b[43m=\u001b[49m\u001b[43mpresence_penalty\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1915\u001b[39m \u001b[43m \u001b[49m\u001b[43mrepeat_penalty\u001b[49m\u001b[43m=\u001b[49m\u001b[43mrepeat_penalty\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1916\u001b[39m \u001b[43m \u001b[49m\u001b[43mtop_k\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtop_k\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1917\u001b[39m \u001b[43m \u001b[49m\u001b[43mstream\u001b[49m\u001b[43m=\u001b[49m\u001b[43mstream\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1918\u001b[39m \u001b[43m \u001b[49m\u001b[43mseed\u001b[49m\u001b[43m=\u001b[49m\u001b[43mseed\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1919\u001b[39m \u001b[43m \u001b[49m\u001b[43mtfs_z\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtfs_z\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1920\u001b[39m \u001b[43m \u001b[49m\u001b[43mmirostat_mode\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmirostat_mode\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1921\u001b[39m \u001b[43m \u001b[49m\u001b[43mmirostat_tau\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmirostat_tau\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1922\u001b[39m \u001b[43m \u001b[49m\u001b[43mmirostat_eta\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmirostat_eta\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1923\u001b[39m \u001b[43m \u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1924\u001b[39m \u001b[43m \u001b[49m\u001b[43mstopping_criteria\u001b[49m\u001b[43m=\u001b[49m\u001b[43mstopping_criteria\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1925\u001b[39m \u001b[43m \u001b[49m\u001b[43mlogits_processor\u001b[49m\u001b[43m=\u001b[49m\u001b[43mlogits_processor\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1926\u001b[39m \u001b[43m \u001b[49m\u001b[43mgrammar\u001b[49m\u001b[43m=\u001b[49m\u001b[43mgrammar\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1927\u001b[39m \u001b[43m \u001b[49m\u001b[43mlogit_bias\u001b[49m\u001b[43m=\u001b[49m\u001b[43mlogit_bias\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 1928\u001b[39m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n", "\u001b[36mFile \u001b[39m\u001b[32m~/Codes/trans/.venv/lib/python3.11/site-packages/llama_cpp/llama.py:1835\u001b[39m, in \u001b[36mLlama.create_completion\u001b[39m\u001b[34m(self, prompt, suffix, max_tokens, temperature, top_p, min_p, typical_p, logprobs, echo, stop, frequency_penalty, presence_penalty, repeat_penalty, top_k, stream, seed, tfs_z, mirostat_mode, mirostat_tau, mirostat_eta, model, stopping_criteria, logits_processor, grammar, logit_bias)\u001b[39m\n\u001b[32m 1833\u001b[39m chunks: Iterator[CreateCompletionStreamResponse] = completion_or_chunks\n\u001b[32m 1834\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m chunks\n\u001b[32m-> \u001b[39m\u001b[32m1835\u001b[39m completion: Completion = \u001b[38;5;28;43mnext\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mcompletion_or_chunks\u001b[49m\u001b[43m)\u001b[49m \u001b[38;5;66;03m# type: ignore\u001b[39;00m\n\u001b[32m 1836\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m completion\n", "\u001b[36mFile \u001b[39m\u001b[32m~/Codes/trans/.venv/lib/python3.11/site-packages/llama_cpp/llama.py:1275\u001b[39m, in \u001b[36mLlama._create_completion\u001b[39m\u001b[34m(self, prompt, suffix, max_tokens, temperature, top_p, min_p, typical_p, logprobs, echo, stop, frequency_penalty, presence_penalty, repeat_penalty, top_k, stream, seed, tfs_z, mirostat_mode, mirostat_tau, mirostat_eta, model, stopping_criteria, logits_processor, grammar, logit_bias)\u001b[39m\n\u001b[32m 1270\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mlen\u001b[39m(prompt_tokens) >= \u001b[38;5;28mself\u001b[39m._n_ctx:\n\u001b[32m 1271\u001b[39m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mValueError\u001b[39;00m(\n\u001b[32m 1272\u001b[39m \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mRequested tokens (\u001b[39m\u001b[38;5;132;01m{\u001b[39;00m\u001b[38;5;28mlen\u001b[39m(prompt_tokens)\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m) exceed context window of \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mllama_cpp.llama_n_ctx(\u001b[38;5;28mself\u001b[39m.ctx)\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m\"\u001b[39m\n\u001b[32m 1273\u001b[39m )\n\u001b[32m-> \u001b[39m\u001b[32m1275\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m max_tokens \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m \u001b[38;5;129;01mor\u001b[39;00m \u001b[43mmax_tokens\u001b[49m\u001b[43m \u001b[49m\u001b[43m<\u001b[49m\u001b[43m=\u001b[49m\u001b[43m \u001b[49m\u001b[32;43m0\u001b[39;49m:\n\u001b[32m 1276\u001b[39m \u001b[38;5;66;03m# Unlimited, depending on n_ctx.\u001b[39;00m\n\u001b[32m 1277\u001b[39m max_tokens = \u001b[38;5;28mself\u001b[39m._n_ctx - \u001b[38;5;28mlen\u001b[39m(prompt_tokens)\n\u001b[32m 1279\u001b[39m \u001b[38;5;66;03m# Truncate max_tokens if requested tokens would exceed the context window\u001b[39;00m\n", "\u001b[31mTypeError\u001b[39m: '<=' not supported between instances of 'str' and 'int'" ] } ], "source": [ "translator(\"欢迎来到go sim大会\", \"zh\", 'en')" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": ".venv", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.11.11" } }, "nbformat": 4, "nbformat_minor": 2 }