Closed Trailingslashes closed 4 weeks ago
sking to truncate to max_length but no maximum length is provided and the model has no predefined maximum length. Default to no truncation.
Thanks, will take a look. There is a huge range of models on Hugging Face and some have quite different APIs from others.
Two things to try:
model_type
of huggingface.model
AttributeError: module 'garak.generators.huggingface' has no attribute 'model'. Did you mean: 'Model'? (garak) ✘ $ ~/garak garak --model_type huggingface.Model --model_name "google/gemma-7b" --probes all --detectors all --parallel_requests 5 --parallel_attempts 5 garak LLM security probe v0.9.0.12 ( https://github.com/leondz/garak ) at 2024-03-14T09:29:07.085797 📜 reporting to garak_runs/garak.bfb1e889-0983-4a80-8fce-89d5646e2ce3.report.jsonl 🦜 loading generator: Hugging Face 🤗 model: google/gemma-7b Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4/4 [00:49<00:00, 12.46s/it] 🕵️ queue of probes: atkgen.Tox, continuation.ContinueSlursReclaimedSlurs80, dan.AntiDAN, dan.AutoDANProbe, dan.ChatGPT_Developer_Mode_RANTI, dan.ChatGPT_Developer_Mode_v2, dan.ChatGPT_Image_Markdown, dan.DAN_Jailbreak, dan.DUDE, dan.Dan_10_0, dan.Dan_11_0, dan.Dan_6_0, dan.Dan_6_2, dan.Dan_7_0, dan.Dan_8_0, dan.Dan_9_0, dan.STAN, encoding.InjectAscii85, encoding.InjectBase16, encoding.InjectBase2048, encoding.InjectBase32, encoding.InjectBase64, encoding.InjectBraille, encoding.InjectEcoji, encoding.InjectHex, encoding.InjectMorse, encoding.InjectNato, encoding.InjectROT13, encoding.InjectUU, encoding.InjectZalgo, gcg.GCG_Probe, glitch.Glitch100, goodside.Davidjl, goodside.Tag, goodside.ThreatenJSON, goodside.WhoIsRiley, knownbadsignatures.EICAR, knownbadsignatures.GTUBE, knownbadsignatures.GTphish, leakreplay.GuardianCloze, leakreplay.GuardianComplete, leakreplay.LiteratureCloze80, leakreplay.LiteratureComplete80, leakreplay.NYTCloze, leakreplay.NYTComplete, lmrc.Anthropomorphisation, lmrc.Bullying, lmrc.Deadnaming, lmrc.Profanity, lmrc.QuackMedicine, lmrc.SexualContent, lmrc.Sexualisation, lmrc.SlurUsage, malwaregen.Evasion, malwaregen.Payload, malwaregen.SubFunctions, malwaregen.TopLevel, misleading.FalseAssertion50, packagehallucination.Python, promptinject.HijackHateHumansMini, promptinject.HijackKillHumansMini, promptinject.HijackLongPromptMini, realtoxicityprompts.RTPBlank, realtoxicityprompts.RTPFlirtation, realtoxicityprompts.RTPIdentity_Attack, realtoxicityprompts.RTPInsult, realtoxicityprompts.RTPProfanity, realtoxicityprompts.RTPSevere_Toxicity, realtoxicityprompts.RTPSexually_Explicit, realtoxicityprompts.RTPThreat, replay.Repeat, snowball.GraphConnectivityMini, snowball.PrimesMini, snowball.SenatorsMini, tap.TAPCached, xss.MarkdownImageExfil 🔎 queue of detectors: always.Fail, always.Pass, continuation.Continuation, dan.AntiDAN, dan.DAN, dan.DANJailbreak, dan.DUDE, dan.DevMode, dan.MarkdownLink, dan.STAN, encoding.DecodeApprox, encoding.DecodeMatch, goodside.Glitch, goodside.PlainJSON, goodside.RileyIsnt, knownbadsignatures.EICAR, knownbadsignatures.GTUBE, knownbadsignatures.GTphish, leakreplay.StartsWith, lmrc.Anthro, lmrc.QuackMedicine, malwaregen.AnyCode, misleading.MustRefuteClaimKW, misleading.MustRefuteClaimModel, mitigation.MitigationBypass, packagehallucination.PythonPypi, promptinject.AttackRogueString, replay.RepeatDiverges, riskywords.LDNOOBW, riskywords.OfcomOffensiveDiscriminatory, riskywords.OfcomOffensiveGeneral, riskywords.OfcomOffensiveMentalHealth, riskywords.OfcomOffensiveRaceEthnic, riskywords.OfcomOffensiveSexGender, riskywords.OfcomOffensiveSexual, riskywords.SurgeProfanityAnimalReferences, riskywords.SurgeProfanityBodilyFluids, riskywords.SurgeProfanityMentalDisability, riskywords.SurgeProfanityPhysicalAttributes, riskywords.SurgeProfanityPhysicalDisability, riskywords.SurgeProfanityPolitical, riskywords.SurgeProfanityRacialEthnic, riskywords.SurgeProfanityReligious, riskywords.SurgeProfanitySexual, riskywords.SurgeProfanitySexualOrientationGender, snowball.DontStartNo, snowball.DontStartYes, specialwords.SlursReclaimedSlurs, toxicity.ToxicCommentModel, xss.MarkdownExfilBasic, xss.MarkdownExfilContent 🔴🪖 🦜 loading generator: Hugging Face 🤗 pipeline: leondz/artgpt2tox probes.atkgen.Tox: 0%| | 0/10 [00:00<?, ?it/sAsking to truncate to max_length but no maximum length is provided and the model has no predefined maximum length. Default to no truncation.
@leondz, no luck, but i'll give NVCF a shot.
Python3.12 garak --version garak LLM security probe v0.9.0.12 ( https://github.com/leondz/garak ) at 2024-03-13T16:54:09.098209
Running any model from garak results in an error: