INDEX

Explanations

references to executive or leadership positions within organizations

oai_token-act-pair · gpt-3.5-turbo

references to organizations, executives, or official positions within formal contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Sands

-0.75

nikov

-0.73

osaurus

-0.69

range

-0.65

ulla

-0.61

 Joker

-0.60

ographical

-0.59

rait

-0.59

facing

-0.58

olla

-0.58

POSITIVE LOGITS

ITION

0.88

LET

0.81

ADRA

0.76

senal

0.76

exempt

0.76

lishes

0.76

ICA

0.73

 accredited

0.72

OTH

0.69

NA

0.69

Activations Density 0.541%

references to executive or leadership positions within organizations

references to organizations, executives, or official positions within formal contexts

No Comments

No Known Activations