INDEX

Explanations

positive adjectives expressing approval

oai_token-act-pair · gpt-3.5-turbo

phrases related to the impact of actions on political or economic outcomes

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

ricted

-0.71

 ALWAYS

-0.68

 virginity

-0.68

 constantly

-0.67

 mummy

-0.65

=""

-0.64

 backlog

-0.63

 WHERE

-0.63

yne

-0.62

 continually

-0.62

POSITIVE LOGITS

 cheered

0.97

 rhetorical

0.96

 prophetic

0.94

 applauded

0.93

 ausp

0.91

 ironic

0.91

 unexpected

0.90

 populist

0.89

 bullish

0.87

 triumphant

0.87

Activations Density 0.447%

positive adjectives expressing approval

phrases related to the impact of actions on political or economic outcomes

No Comments

No Known Activations