INDEX

Explanations

career-related achievements expressed with superlatives, such as "high" and "best."

oai_token-act-pair · gpt-3.5-turbo

phrases related to career achievements and performance metrics

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

Square

-0.70

src

-0.68

umo

-0.67

mass

-0.66

anmar

-0.66

forums

-0.66

Pros

-0.66

microsoft

-0.64

mop

-0.64

Night

-0.64

POSITIVE LOGITS

 scorer

0.92

 holder

0.76

 goaltender

0.75

LEASE

0.75

 hitters

0.75

 holders

0.70

 breaker

0.70

 record

0.70

inel

0.69

 designation

0.67

Activations Density 0.113%

career-related achievements expressed with superlatives, such as "high" and "best."

phrases related to career achievements and performance metrics

No Comments

No Known Activations