INDEX

Explanations

numbers surrounded by certain characters, possibly indicating specific data points or references

oai_token-act-pair · gpt-3.5-turbo

sequences of numerical codes or identifiers

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

akening

-0.91

ĸļ

-0.88

raints

-0.80

ategory

-0.80

ItemTracker

-0.79

aken

-0.76

 Seym

-0.76

rolet

-0.75

=-=-

-0.74

iless

-0.74

POSITIVE LOGITS

LECT

0.87

0.82

rd

0.82

inately

0.77

aan

0.75

Â¯Â¯Â¯Â¯

0.71

illard

0.70

0.69

 deficient

0.68

ptin

0.66

Activations Density 0.032%

numbers surrounded by certain characters, possibly indicating specific data points or references

sequences of numerical codes or identifiers

No Comments

No Known Activations