<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[各位大神有试过qwen3.8 27b超过256k 上下文怎么样吗?]]></title><description><![CDATA[<p dir="auto">我有双卡r9700不太需要多会话。所以想问一下有人过超过256k的效果怎么样？</p>
]]></description><link>https://lcz.me/topic/3119</link><generator>RSS for Node</generator><lastBuildDate>Thu, 19 Nov 2026 00:16:50 GMT</lastBuildDate><atom:link href="https://lcz.me/topic/3119.rss" rel="self" type="application/rss+xml"/><pubDate>Mon, 16 Nov 2026 09:01:59 GMT</pubDate><ttl>60</ttl><item><title><![CDATA[Reply to 各位大神有试过qwen3.8 27b超过256k 上下文怎么样吗? on Tue, 17 Nov 2026 03:12:53 GMT]]></title><description><![CDATA[<p dir="auto">主要看你的硬件，本质 不就是猜token机器，上下文越长，轮次越多，噪声越多，F16原版也没办法稳定不出错。 Q8就更加重 语义偏离了。</p>
]]></description><link>https://lcz.me/post/34029</link><guid isPermaLink="true">https://lcz.me/post/34029</guid><dc:creator><![CDATA[stxpnet]]></dc:creator><pubDate>Tue, 17 Nov 2026 03:12:53 GMT</pubDate></item><item><title><![CDATA[Reply to 各位大神有试过qwen3.8 27b超过256k 上下文怎么样吗? on Mon, 16 Nov 2026 14:29:17 GMT]]></title><description><![CDATA[<p dir="auto"><a class="plugin-mentions-user plugin-mentions-a" href="/user/applejuice" aria-label="Profile: applejuice">@<bdi>applejuice</bdi></a> QFN的测试刚做了一半，三档256/512/1024，目前的结果跟你说的一样，512无下滑，1024下滑了37%。因为一般agent的压缩阈值是0.8，所以准备再测下800K如何。</p>
]]></description><link>https://lcz.me/post/33860</link><guid isPermaLink="true">https://lcz.me/post/33860</guid><dc:creator><![CDATA[neo]]></dc:creator><pubDate>Mon, 16 Nov 2026 14:29:17 GMT</pubDate></item><item><title><![CDATA[Reply to 各位大神有试过qwen3.8 27b超过256k 上下文怎么样吗? on Mon, 16 Nov 2026 14:03:48 GMT]]></title><description><![CDATA[<p dir="auto"><a class="plugin-mentions-user plugin-mentions-a" href="/user/neo" aria-label="Profile: neo">@<bdi>neo</bdi></a> 我没用27b 超过256k 因为实在太慢了</p>
<p dir="auto">但是我用 qwen next flash 512k 成绩跟256k 没分别</p>
<p dir="auto">有兴趣可以叫agent 试试</p>
<pre><code>#!/usr/bin/env python3
"""long_suite.py - harder long-context checks than a single needle, through a running Strata server.

For each length one document is built (the same filler as Strata's tools/needle_bench.py) with every test item planted
in it at fixed depths; the questions are then asked one after another with the same document as prefix, so Strata
reads the document once and only the question after that. Thinking off, greedy, short answers. The document and the
items depend only on the length (fixed seed), so two server configs get identical tests.

Tasks
  multikey  8 people each get a similar code word (word-word-123); 4 of them are asked for       (confusable facts)
  vartrack  two chains of 4 assignments (VAR_x = number, VAR_y = VAR_x, ...) spread over the text;
            the last variable of each chain is asked for                                         (multi-hop)
  nolima    3 facts phrased without the question's words (lives next to the Semper Opera House -&gt;
            "who has been to Dresden?")                                                         (no literal match)

    python3 long_suite.py --lengths 64k,128k,256k --label cfg262k --out results.json
"""
import argparse, json, random, sys, time, urllib.request
from pathlib import Path

sys.path.insert(0, str(Path.home() / "bench_haystack" / "tools"))   # was ~/strata-hip/tools (deleted 10-26): a copy of the image /opt/strata text folders (haystack source)
import needle_bench as NB  # haystack() + WORDS

NAMES = ["Marisol Venn", "Torsten Achebe", "Ilse Okafor", "Ravi Lindqvist", "Noor Castellanos", "Bram Takahashi",
         "Celia Oduya", "Henrik Moreau"]
VARS = ["VAR_KQX", "VAR_PLM", "VAR_ZTR", "VAR_WEN", "VAR_HUD", "VAR_BOC", "VAR_JYF", "VAR_SAG"]
NOLIMA = [("Yuki", "Actually, Yuki lives right next to the Semper Opera House.", "Dresden"),
          ("Dmitri", "As it happens, Dmitri spent last summer walking the gardens of the Alhambra every morning.", "Granada"),
          ("Amara", "Amara's office window looks straight at the Petronas Twin Towers.", "Kuala Lumpur")]


def build(tokens: int):
    rnd = random.Random(tokens)
    text = NB.haystack(int(tokens * 0.97 * NB.CHARS_PER_TOKEN))
    items, plants = [], []                       # plants: (depth %, sentence)
    # multikey: 8 codes at depths 5..95, ask 4 of them
    depths = [5, 17, 29, 41, 53, 65, 77, 95]
    codes = {}
    for name, d in zip(NAMES, depths):
        code = f"{rnd.choice(NB.WORDS)}-{rnd.choice(NB.WORDS)}-{rnd.randint(100, 999)}"
        codes[name] = (code, d)
        plants.append((d, f"Registry entry: the special access code assigned to {name} is {code}."))
    for name in (NAMES[0], NAMES[3], NAMES[5], NAMES[7]):
        code, d = codes[name]
        items.append(("multikey", d, f"In the text above, what is the special access code assigned to {name}? "
                      "Reply with the code only.", code))
    # vartrack: two interleaved 4-hop chains
    chains = [(VARS[0:4], [8, 31, 58, 86]), (VARS[4:8], [14, 44, 71, 92])]
    for vs, ds in chains:
        val = str(rnd.randint(10000, 99999))
        plants.append((ds[0], f"Config note: let {vs[0]} = {val}."))
        for prev, cur, d in zip(vs, vs[1:], ds[1:]):
            plants.append((d, f"Config note: let {cur} = {prev}."))
        items.append(("vartrack", ds[0], f"Following the 'Config note: let' assignments in the text above, which number "
                      f"does {vs[-1]} ultimately equal? Reply with the number only.", val))
    # nolima
    for (name, sent, city), d in zip(NOLIMA, [20, 55, 90]):
        plants.append((d, sent))
        items.append(("nolima", d, f"Based on the text above, which character has been to {city}? "
                      "Reply with the name only.", name))
    # plant at line starts, from the deepest so earlier offsets stay valid
    for d, sent in sorted(plants, key=lambda p: -p[0]):
        cut = int(len(text) * d / 100)
        cut = text.rfind("\n", 0, cut) + 1 or cut
        text = text[:cut] + "\n" + sent + "\n" + text[cut:]
    return text, items


def ask(url, doc, q, timeout):
    body = {"model": "strata", "max_tokens": 60, "temperature": 0, "chat_template_kwargs": {"enable_thinking": False},
            "messages": [{"role": "user", "content": doc + "\n\n" + q}]}
    req = urllib.request.Request(url.rstrip("/") + "/v1/chat/completions", data=json.dumps(body).encode(),
                                 headers={"Content-Type": "application/json"})
    t0 = time.time()
    with urllib.request.urlopen(req, timeout=timeout) as r:
        out = json.loads(r.read())
    return (out["choices"][0]["message"].get("content") or "").strip(), out["usage"]["prompt_tokens"], time.time() - t0


def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--url", default="http://127.0.0.1:8080")
    ap.add_argument("--lengths", default="64k,128k")
    ap.add_argument("--label", default="")
    ap.add_argument("--timeout", type=float, default=3600)
    ap.add_argument("--out")
    a = ap.parse_args()
    rows = []
    for L in a.lengths.split(","):
        tokens = int(float(L.lower().rstrip("k")) * 1024) if L.lower().endswith("k") else int(L)
        doc, items = build(tokens)
        for task, d, q, want in items:
            try:
                ans, n, secs = ask(a.url, doc, q, a.timeout)
            except Exception as e:
                print(f"{a.label} {L:&gt;5} {task:9} d{d:&gt;3}%: request failed ({e})", flush=True)
                rows.append({"label": a.label, "length": L, "task": task, "depth": d, "error": str(e)})
                continue
            ok = want.lower() in ans.lower()
            print(f"{a.label} {L:&gt;5} {task:9} d{d:&gt;3}%: {'OK  ' if ok else 'MISS'} ({n:,} tok, {secs:5.0f} s)"
                  + ("" if ok else f"  want {want!r} got {ans[:70]!r}"), flush=True)
            rows.append({"label": a.label, "length": L, "task": task, "depth": d, "ok": ok, "want": want,
                         "answer": ans[:200], "prompt_tokens": n, "seconds": round(secs, 1)})
        done = [r for r in rows if r["length"] == L and "ok" in r]
        by = {t: f"{sum(r['ok'] for r in done if r['task'] == t)}/{sum(1 for r in done if r['task'] == t)}"
              for t in ("multikey", "vartrack", "nolima")}
        print(f"{a.label} {L:&gt;5} SUMMARY  multikey {by['multikey']}  vartrack {by['vartrack']}  nolima {by['nolima']}"
              f"  total {sum(r['ok'] for r in done)}/{len(done)}", flush=True)
    if a.out:
        Path(a.out).write_text(json.dumps(rows, indent=1), encoding="utf-8")


if __name__ == "__main__":
    main()

</code></pre>
]]></description><link>https://lcz.me/post/33846</link><guid isPermaLink="true">https://lcz.me/post/33846</guid><dc:creator><![CDATA[applejuice]]></dc:creator><pubDate>Mon, 16 Nov 2026 14:03:48 GMT</pubDate></item><item><title><![CDATA[Reply to 各位大神有试过qwen3.8 27b超过256k 上下文怎么样吗? on Mon, 16 Nov 2026 13:14:40 GMT]]></title><description><![CDATA[<p dir="auto">128k以后会打醉拳螳螂拳虎鹤双型什么的，其实是64k就开始语义化整理出现偏差</p>
]]></description><link>https://lcz.me/post/33839</link><guid isPermaLink="true">https://lcz.me/post/33839</guid><dc:creator><![CDATA[范志刚]]></dc:creator><pubDate>Mon, 16 Nov 2026 13:14:40 GMT</pubDate></item><item><title><![CDATA[Reply to 各位大神有试过qwen3.8 27b超过256k 上下文怎么样吗? on Mon, 16 Nov 2026 12:57:23 GMT]]></title><description><![CDATA[<p dir="auto">我做过测试350k，单针跟256k一样很准，但多针命中率下降得厉害。</p>
]]></description><link>https://lcz.me/post/33836</link><guid isPermaLink="true">https://lcz.me/post/33836</guid><dc:creator><![CDATA[neo]]></dc:creator><pubDate>Mon, 16 Nov 2026 12:57:23 GMT</pubDate></item><item><title><![CDATA[Reply to 各位大神有试过qwen3.8 27b超过256k 上下文怎么样吗? on Mon, 16 Nov 2026 11:20:53 GMT]]></title><description><![CDATA[<p dir="auto">qwen 4 有可能。而且快公布了。等等吧。</p>
]]></description><link>https://lcz.me/post/33804</link><guid isPermaLink="true">https://lcz.me/post/33804</guid><dc:creator><![CDATA[williamlouis]]></dc:creator><pubDate>Mon, 16 Nov 2026 11:20:53 GMT</pubDate></item><item><title><![CDATA[Reply to 各位大神有试过qwen3.8 27b超过256k 上下文怎么样吗? on Mon, 16 Nov 2026 11:18:54 GMT]]></title><description><![CDATA[<p dir="auto">Qwen3.8 27B的上下文最大 256K, 到一定阶段(Hermes缺省50%), 就会启动压缩, 以减少上下文.  长链任务和多轮对话, 很有可能会突破256K, 但我实测本地27B NVFP4的表现不错, 基本不会跑偏.</p>
]]></description><link>https://lcz.me/post/33802</link><guid isPermaLink="true">https://lcz.me/post/33802</guid><dc:creator><![CDATA[Tony Wang]]></dc:creator><pubDate>Mon, 16 Nov 2026 11:18:54 GMT</pubDate></item><item><title><![CDATA[Reply to 各位大神有试过qwen3.8 27b超过256k 上下文怎么样吗? on Mon, 16 Nov 2026 09:40:10 GMT]]></title><description><![CDATA[<p dir="auto">160K 以上會錯亂，不穩定（我沒證據，只憑感覺）<br />
所以我都設128K+subprocess工作(機器夠快，已經夠用了)</p>
]]></description><link>https://lcz.me/post/33779</link><guid isPermaLink="true">https://lcz.me/post/33779</guid><dc:creator><![CDATA[David Chen]]></dc:creator><pubDate>Mon, 16 Nov 2026 09:40:10 GMT</pubDate></item></channel></rss>