INDEX

Explanations

direct quotes from "reporters" in different political contexts

oai_token-act-pair · gpt-3.5-turbo

mentions of "reporters."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ateral

-0.75

ston

-0.73

emonic

-0.72

erton

-0.71

vasive

-0.70

inence

-0.67

uras

-0.65

qual

-0.62

duc

-0.60

tz

-0.60

POSITIVE LOGITS

 reporters

0.93

 reporter

0.85

orters

0.80

cov

0.78

etter

0.77

 journalists

0.77

 Journalists

0.74

verage

0.74

hip

0.74

 RELEASE

0.73

Activations Density 0.019%

direct quotes from "reporters" in different political contexts

mentions of "reporters."

No Comments

No Known Activations