Use MediaTek Research’s multimodal Traditional Chinese model for Taiwanese knowledge, visual instructions, conversation, and function calling. Use the playground to give Llama Breeze2 8B Instruct a representative input and review its output. Adjust the available settings to shape the result for your task.
Model
MediaTek-Research/Llama-Breeze2-8B-Instruct
Documentation
Use this reference to understand how every field in the playground changes the request sent to MediaTek-Research/Llama-Breeze2-8B-Instruct.
Required
Type: JSON array
Provides the chronological conversation as a JSON array of role-and-content messages. Content may be plain text or multimodal items; place the user instruction and any image reference in the same turn so the prompt processor can associate them correctly.
Optional
Type: File
Default: None
Uploads an optional image for visual instruction following alongside the conversation. Leave it empty for text-only chat; when supplied, make the conversation explicitly refer to the image and use a clear source with enough resolution for the requested details.
Optional
Type: JSON array
Default: None
Defines optional callable functions as JSON objects containing names, descriptions, and parameter schemas. The model may request a matching function, but the application must execute it and return the result; definitions should be precise enough for reliable argument generation.
Optional
Type: Integer
Default: 2048
Range: 1–131072
Caps how many tokens the model can add after the input. The model may stop earlier naturally, but a limit that is too low can cut off an answer; input tokens are not counted in this value.
Optional
Type: Boolean
Default: On
Chooses whether the next token is sampled from a probability distribution or selected deterministically. Enable it for more varied responses that are affected by temperature and Top P; disable it when repeatability matters or when using beam search.
Optional
Type: Number
Default: 0.01
Range: 0–2
Adjusts how strongly the model favors its most likely next token. Lower values produce more focused and repeatable output, while higher values allow less-likely choices and increase variety; it only has an effect when sampling is enabled.
Optional
Type: Number
Default: 0.01
Range: 0–1
Limits sampling to the smallest set of likely tokens whose probabilities add up to this value. Lower settings make output more focused, while a value near 1 keeps more alternatives available; it only affects sampled generation.
Optional
Type: Number
Default: 1.1
Range: 0.01–Unbounded
Changes the scores of tokens that already appeared in the generated text. Values above 1 discourage repeated words and loops, 1 leaves probabilities unchanged, and values below 1 make repetition more likely.
Optional
Type: Integer
Default: 128009
Range: 0–Unbounded
Sets the numeric end-of-sequence token that tells generation when a response is complete. Keep the model’s documented default unless using a compatible custom chat template, because an incorrect ID can prevent stopping or terminate output prematurely.
Optional
Type: JSON object
Default: None
Passes additional JSON overrides to Transformers GenerationConfig. Use it for supported generation controls not exposed separately, and avoid duplicate keys because they may conflict with or override values elsewhere in the form.