Oratis commited on
Commit
7eeffa9
·
verified ·
1 Parent(s): 4c6734f

更新论文组织名称与资料链接

Browse files

使首页组织名称和链接与当前资料页一致,保留作者署名与研究正文。已核对 14 页:正文渲染保持一致,旧组织文本及链接已移除。README 记录本次展示信息更新。

Files changed (2) hide show
  1. README.md +2 -0
  2. main.pdf +2 -2
README.md CHANGED
@@ -16,6 +16,8 @@ pretty_name: "Externalizing the Workspace"
16
 
17
  Language-model agents drift. Building on Anthropic's finding that LLMs maintain a **verbalizable global workspace** (Gurnee et al., 2026), this paper models persistent self-state agent architectures (LISA's *soul*) as **functional externalizations of the workspace** — and backs the model with experiments that run entirely on open models and independent-researcher compute.
18
 
 
 
19
  ## Headline results
20
 
21
  | finding | evidence |
 
16
 
17
  Language-model agents drift. Building on Anthropic's finding that LLMs maintain a **verbalizable global workspace** (Gurnee et al., 2026), this paper models persistent self-state agent architectures (LISA's *soul*) as **functional externalizations of the workspace** — and backs the model with experiments that run entirely on open models and independent-researcher compute.
18
 
19
+ > Presentation update (2026-10-16): the PDF affiliation label and organization profile link now match this repository’s current name. The author byline and research text are unchanged.
20
+
21
  ## Headline results
22
 
23
  | finding | evidence |
main.pdf CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8fb0bb99d4f10baacece4c9c9c5171dabe8338e95b12adaa584063911395bc87
3
- size 715926
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2ea9b35c7da0c2016de2027777aa7e31494260dee5af1f47feb2c79d647de289
3
+ size 747204