INDEX

Explanations

dates or time-related terms

oai_token-act-pair · gpt-3.5-turbo

references to time-related actions and progress in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĥ´ãĤ¡

-0.66

ceivable

-0.66

 Fame

-0.58

 airs

-0.58

 Nightmares

-0.58

 legend

-0.58

 Cthulhu

-0.57

 SOME

-0.56

 Atlantis

-0.56

 HUGE

-0.55

POSITIVE LOGITS

 combating

1.34

 enforcing

1.15

 prosecuting

1.14

 resolving

1.14

 implementing

1.13

 negotiating

1.12

 regulating

1.10

 administering

1.09

 pursuing

1.08

 tackling

1.08

Activations Density 0.551%

dates or time-related terms

references to time-related actions and progress in various contexts

No Comments

No Known Activations