INDEX

Explanations

phrases related to responses or reactions from various sources or individuals

oai_token-act-pair · gpt-3.5-turbo

instances of the word "responded."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

cutting

-0.83

stained

-0.75

marks

-0.72

our

-0.71

fall

-0.70

thin

-0.69

clad

-0.68

raft

-0.68

free

-0.66

 illusion

-0.66

POSITIVE LOGITS

 responses

0.93

 responds

0.91

 respond

0.90

 responded

0.87

 replies

0.87

umph

0.84

 thereto

0.83

reply

0.81

 receptive

0.80

isson

0.80

Activations Density 0.015%

phrases related to responses or reactions from various sources or individuals

instances of the word "responded."

No Comments

No Known Activations