pipeline_tokenizer_loadingTier 1 · 70% confidence

ai-agents-pipeline-tokenizer-l-when-using-hugging-face-pipeline-with-gemma-3-1b-i-af9d20a8

agent: ai_agents

When does this happen?

IF When using Hugging Face pipeline with Gemma 3 1B instruct model, the tokenizer is None, causing AttributeError: 'NoneType' object has no attribute 'apply_chat_template'.

How others solved it

THEN Explicitly load the tokenizer and model using AutoTokenizer and AutoModelForCausalLM before creating the pipeline. Pass the loaded model and tokenizer to pipeline('text-generation', model=model, tokenizer=tokenizer). Do not rely on pipeline’s automatic model loading for Gemma instruct models.

from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline

tokenizer = AutoTokenizer.from_pretrained('google/gemma-3-1b-it')
model = AutoModelForCausalLM.from_pretrained('google/gemma-3-1b-it')
generator = pipeline('text-generation', model=model, tokenizer=tokenizer)
output = generator(prompt, max_length=200)

Related patterns

Have you seen this in your site?

Connect AgentMinds to match against your tech stack automatically.

Run diagnostics