INDEX

Explanations

unexpected or ironic situations and events

oai_token-act-pair · gpt-3.5-turbo

instances of irony or interesting contrasts in the text

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

 comprehension

-0.72

 gulf

-0.65

 commitment

-0.65

 sewage

-0.62

division

-0.61

verbal

-0.60

"},"

-0.58

 Submission

-0.58

 kindred

-0.58

iens

-0.58

POSITIVE LOGITS

ffe

0.71

 situated

0.71

 coinc

0.68

 ironically

0.67

 Ironically

0.67

haus

0.67

 titled

0.66

identally

0.65

leck

0.64

pmwiki

0.63

Activations Density 0.030%

unexpected or ironic situations and events

instances of irony or interesting contrasts in the text

No Comments

No Known Activations