Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -85,7 +85,7 @@ The current value of Key1 is Value_N.
|
|
85 |
|
86 |
|
87 |
## Results:
|
88 |
-
LLMs **cannot reliably retrieve** Value_N. Distribution spans value_1 to value_N, and **as N increases**, the **answers skew** increasingly toward **value_1**.
|
89 |
|
90 |
|
91 |
|
@@ -114,7 +114,7 @@ Our test adopts the **classic proactive** interference paradigm from cognitive s
|
|
114 |
|
115 |
See: https://sites.google.com/view/cog4llm
|
116 |
|
117 |
-
##
|
118 |
- Humans: near-ceiling accuracy (99%+) on this controlled task across conditions (see paper for protocol and exact numbers).
|
119 |
- LLMs: accuracy declines approximately log-linearly with the number of updates per key and with the number of concurrent update blocks (details, plots, and model list in our paper).
|
120 |
|
|
|
85 |
|
86 |
|
87 |
## Results:
|
88 |
+
ALL tested SOTA LLMs **cannot reliably retrieve** Value_N. Distribution spans value_1 to value_N, and **as N increases**, the **answers skew** increasingly toward **value_1**.
|
89 |
|
90 |
|
91 |
|
|
|
114 |
|
115 |
See: https://sites.google.com/view/cog4llm
|
116 |
|
117 |
+
## SAME Log-linear Decline of Accuracy for ALL SOTA LLMs tested(2019-2025)
|
118 |
- Humans: near-ceiling accuracy (99%+) on this controlled task across conditions (see paper for protocol and exact numbers).
|
119 |
- LLMs: accuracy declines approximately log-linearly with the number of updates per key and with the number of concurrent update blocks (details, plots, and model list in our paper).
|
120 |
|