INDEX

Explanations

words related to uncovering or unveiling information

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "reveals."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 peaceful

-0.64

hare

-0.62

Handle

-0.62

 safe

-0.60

 automatic

-0.60

 reserve

-0.60

aterasu

-0.60

 retired

-0.59

 handle

-0.57

 blanket

-0.57

POSITIVE LOGITS

 reveals

3.13

 confirms

2.06

 exposes

1.71

 shows

1.68

 demonstrates

1.66

 tells

1.63

 suggests

1.62

 discl

1.61

 proves

1.60

 explains

1.59

Activations Density 0.020%

words related to uncovering or unveiling information

occurrences of the word "reveals."

No Comments

No Known Activations