INDEX

Explanations

statements of opinions or reactions to news events

oai_token-act-pair · gpt-3.5-turbo

terms related to decisions, actions, and their consequences in a political context

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

conservancy

-0.68

 ILCS

-0.66

EQU

-0.59

[|

-0.58

anyl

-0.58

"))

-0.58

 aging

-0.57

 ageing

-0.57

 Certified

-0.57

 majorities

-0.54

POSITIVE LOGITS

 saying

0.89

 stating

0.75

 citing

0.75

 remark

0.74

but

0.72

 announcement

0.70

 reiter

0.69

 noting

0.69

 suggestion

0.68

 dismissing

0.67

Activations Density 0.295%

statements of opinions or reactions to news events

terms related to decisions, actions, and their consequences in a political context

No Comments

No Known Activations