text-embeddings-inference
text-embeddings-inference copied to clipboard
Input validation error: `inputs` must have less than 32000 characters. Given: 67337
When using /embed and truncate=true with long inputs, the input is correctly truncated at the token level, but I still get a validation error for inputs longer than 32,000 characters:
{
"error": "Input validation error: `inputs` must have less than 32000 characters. Given: 67337",
"error_type": "Validation"
}
It seems like truncate only affects tokenization and not the actual input length. Is the only solution to manually truncate the input on the client side or is there some parameter that I missed?