☁️ 雲端辦公
所有檔案
分類管理
文件工作區
📝 編輯文件
把儲存操作固定在上方,避免被頁尾或編輯器工具列遮住。
瀏覽預覽
返回列表
刪除檔案
儲存變更
內容編輯區
換一個支援 64K+ 上下文的模型
Document
<p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">這個錯誤很清楚,問題是 <code class="bg-text-200/5 border border-0.5 border-border-300 text-danger-000 whitespace-pre-wrap rounded-[0.4rem] px-1 py-px text-[0.9rem]">qwen2.5:7b</code> 的上下文視窗只有 32,768 tokens,但 Hermes Agent 要求至少 64,000 tokens。</p> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">以下是解決方法:</p> <hr class="border-border-200 border-t-0.5 my-3 mx-1.5"> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]"><strong>方法一 — 換一個支援 64K+ 上下文的模型(推薦)</strong></p> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">先用 Ollama 下載較大的模型:</p> <div class="relative group/copy bg-bg-000/50 border-0.5 border-border-400 rounded-lg focus:outline-none focus-visible:ring-2 focus-visible:ring-accent-100" tabindex="0" role="group" aria-label="bash code"> <div class="sticky opacity-0 group-hover/copy:opacity-100 group-focus-within/copy:opacity-100 top-2 py-2 h-12 w-0 float-right"> <div class="absolute right-0 h-8 px-2 items-center inline-flex z-10"> <div class="relative"> <div class="transition-all opacity-100 scale-100"> </div> <div class="absolute inset-0 flex items-center justify-center"> <div class="transition-all opacity-0 scale-50"> </div> </div> </div> </div> </div> <div class="text-text-500 font-small p-3.5 pb-0">bash</div> <div class="overflow-x-auto"> <pre class="code-block__code !my-0 !rounded-lg !text-sm !leading-relaxed p-3.5"><code class="language-bash">ollama pull llama3.1:8b</code></pre> </div> </div> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">然後修改 <code class="bg-text-200/5 border border-0.5 border-border-300 text-danger-000 whitespace-pre-wrap rounded-[0.4rem] px-1 py-px text-[0.9rem]">config.yaml</code>:</p> <div class="relative group/copy bg-bg-000/50 border-0.5 border-border-400 rounded-lg focus:outline-none focus-visible:ring-2 focus-visible:ring-accent-100" tabindex="0" role="group" aria-label="yaml code"> <div class="sticky opacity-0 group-hover/copy:opacity-100 group-focus-within/copy:opacity-100 top-2 py-2 h-12 w-0 float-right"> <div class="absolute right-0 h-8 px-2 items-center inline-flex z-10"> <div class="relative"> <div class="transition-all opacity-100 scale-100"> </div> <div class="absolute inset-0 flex items-center justify-center"> <div class="transition-all opacity-0 scale-50"> </div> </div> </div> </div> </div> <div class="text-text-500 font-small p-3.5 pb-0">yaml</div> <div class="overflow-x-auto"> <pre class="code-block__code !my-0 !rounded-lg !text-sm !leading-relaxed p-3.5"><code class="language-yaml"><span class="token token key">auxiliary</span><span class="token token">:</span> <span class="token token key">compression</span><span class="token token">:</span> <span class="token token key">model</span><span class="token token">:</span> llama3.1<span class="token token">:</span>8b</code></pre> </div> </div> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">常見支援 64K+ 的模型選擇:</p> <ul class="[li_&]:mb-0 [li_&]:mt-1 [li_&]:gap-1 [&:not(:last-child)_ul]:pb-1 [&:not(:last-child)_ol]:pb-1 list-disc flex flex-col gap-1 pl-8 mb-3"> <li class="whitespace-normal break-words pl-2"><code class="bg-text-200/5 border border-0.5 border-border-300 text-danger-000 whitespace-pre-wrap rounded-[0.4rem] px-1 py-px text-[0.9rem]">llama3.1:8b</code>(128K)</li> <li class="whitespace-normal break-words pl-2"><code class="bg-text-200/5 border border-0.5 border-border-300 text-danger-000 whitespace-pre-wrap rounded-[0.4rem] px-1 py-px text-[0.9rem]">qwen2.5:14b</code>(128K)</li> <li class="whitespace-normal break-words pl-2"><code class="bg-text-200/5 border border-0.5 border-border-300 text-danger-000 whitespace-pre-wrap rounded-[0.4rem] px-1 py-px text-[0.9rem]">mistral-nemo</code>(128K)</li> </ul> <hr class="border-border-200 border-t-0.5 my-3 mx-1.5"> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]"><strong>方法二 — 強制覆蓋上下文長度</strong></p> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">如果您的 <code class="bg-text-200/5 border border-0.5 border-border-300 text-danger-000 whitespace-pre-wrap rounded-[0.4rem] px-1 py-px text-[0.9rem]">qwen2.5:7b</code> 其實已設定更大的 <code class="bg-text-200/5 border border-0.5 border-border-300 text-danger-000 whitespace-pre-wrap rounded-[0.4rem] px-1 py-px text-[0.9rem]">num_ctx</code>,可以直接告訴 Agent 忽略偵測到的值:</p> <div class="relative group/copy bg-bg-000/50 border-0.5 border-border-400 rounded-lg focus:outline-none focus-visible:ring-2 focus-visible:ring-accent-100" tabindex="0" role="group" aria-label="yaml code"> <div class="sticky opacity-0 group-hover/copy:opacity-100 group-focus-within/copy:opacity-100 top-2 py-2 h-12 w-0 float-right"> <div class="absolute right-0 h-8 px-2 items-center inline-flex z-10"> <div class="relative"> <div class="transition-all opacity-100 scale-100"> </div> <div class="absolute inset-0 flex items-center justify-center"> <div class="transition-all opacity-0 scale-50"> </div> </div> </div> </div> </div> <div class="text-text-500 font-small p-3.5 pb-0">yaml</div> <div class="overflow-x-auto"> <pre class="code-block__code !my-0 !rounded-lg !text-sm !leading-relaxed p-3.5"><code class="language-yaml"><span class="token token key">auxiliary</span><span class="token token">:</span> <span class="token token key">compression</span><span class="token token">:</span> <span class="token token key">model</span><span class="token token">:</span> qwen2.5<span class="token token">:</span>7b <span class="token token key">context_length</span><span class="token token">:</span> <span class="token token">65536</span></code></pre> </div> </div> <hr class="border-border-200 border-t-0.5 my-3 mx-1.5"> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]"><strong>方法三 — 在 Ollama 中設定更大的 num_ctx</strong></p> <div class="relative group/copy bg-bg-000/50 border-0.5 border-border-400 rounded-lg focus:outline-none focus-visible:ring-2 focus-visible:ring-accent-100" tabindex="0" role="group" aria-label="bash code"> <div class="sticky opacity-0 group-hover/copy:opacity-100 group-focus-within/copy:opacity-100 top-2 py-2 h-12 w-0 float-right"> <div class="absolute right-0 h-8 px-2 items-center inline-flex z-10"> <div class="relative"> <div class="transition-all opacity-100 scale-100"> </div> <div class="absolute inset-0 flex items-center justify-center"> <div class="transition-all opacity-0 scale-50"> </div> </div> </div> </div> </div> <div class="text-text-500 font-small p-3.5 pb-0">bash</div> <div class="overflow-x-auto"> <pre class="code-block__code !my-0 !rounded-lg !text-sm !leading-relaxed p-3.5"><code class="language-bash">ollama run qwen2.5:7b --num-ctx <span class="token token">65536</span></code></pre> </div> </div> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">或建立一個 Modelfile:</p> <div class="relative group/copy bg-bg-000/50 border-0.5 border-border-400 rounded-lg focus:outline-none focus-visible:ring-2 focus-visible:ring-accent-100" tabindex="0" role="group" aria-label="Code"> <div class="sticky opacity-0 group-hover/copy:opacity-100 group-focus-within/copy:opacity-100 top-2 py-2 h-12 w-0 float-right"> <div class="absolute right-0 h-8 px-2 items-center inline-flex z-10"> <div class="relative"> <div class="transition-all opacity-100 scale-100"> </div> <div class="absolute inset-0 flex items-center justify-center"> <div class="transition-all opacity-0 scale-50"> </div> </div> </div> </div> </div> <div class="overflow-x-auto"> <pre class="code-block__code !my-0 !rounded-lg !text-sm !leading-relaxed p-3.5"><code>FROM qwen2.5:7b PARAMETER num_ctx 65536</code></pre> </div> </div> <hr class="border-border-200 border-t-0.5 my-3 mx-1.5"> <p class="font-claude-response-body break-words whitespace-normal leading-[1.7]">最快的解法是<strong>方法一</strong>,直接換成 <code class="bg-text-200/5 border border-0.5 border-border-300 text-danger-000 whitespace-pre-wrap rounded-[0.4rem] px-1 py-px text-[0.9rem]">llama3.1:8b</code>。請問您知道 <code class="bg-text-200/5 border border-0.5 border-border-300 text-danger-000 whitespace-pre-wrap rounded-[0.4rem] px-1 py-px text-[0.9rem]">config.yaml</code> 在哪個路徑嗎?我可以幫您確認設定。</p>