$ ./build/bin/llama-mtmd-cli -m ~/models/ibm-granite/granite-vision-4.1-4b/granite-4B-vision-4.1-BF16.gguf --mmproj ~/models/ibm-granite/granite-vision-4.1-4b/mmproj-granite-vision-4b-4.1-BF16.gguf --image ./tools/mtmd/test-1.jpeg --temp 0 -n 128 --flash-attn on -p 'what is the publisher name of the newspaper?' 
0.00.038.624 I common_init_result: fitting params to device memory ...
0.00.038.628 I common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on)
0.02.048.601 I common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
0.02.135.834 I mtmd_cli_context: chat template example:
<|start_of_role|>system<|end_of_role|>You are a helpful assistant<|end_of_text|>
<|start_of_role|>user<|end_of_role|>Hello<|end_of_text|>
<|start_of_role|>assistant<|end_of_role|>Hi there<|end_of_text|>
<|start_of_role|>user<|end_of_role|>How are you?<|end_of_text|>
<|start_of_role|>assistant<|end_of_role|>
0.11.508.194 I main: loading model: /Users/ghart/models/ibm-granite/granite-vision-4.1-4b/granite-4B-vision-4.1-BF16.gguf
0.11.508.201 W WARN: This is an experimental CLI for testing multimodal capability.
0.11.508.201 W       For normal use cases, please use the standard llama-cli

new york times


0.13.858.244 W ~llama_context:       MTL0 compute buffer size of 381.5886 MiB, does not match expectation of 226.0117 MiB

$ ./build/bin/llama-mtmd-cli -m ~/models/ibm-granite/granite-speech-4.1-2b-GGUF/granite-speech-4.1-2b-Q4_K_M.gguf --mmproj ~/models/ibm-granite/granite-speech-4.1-2b-GGUF/mmproj-model-f16.gguf  --temp 0 -n 128 --flash-attn on --audio ~/models/ibm-granite/granite-speech-4.1-2b/multilingual_sample.wav -p "transcribe this audio to a text format" --jinja
0.00.039.914 I common_init_result: fitting params to device memory ...
0.00.039.919 I common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on)
0.01.082.039 I common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
0.01.113.404 I mtmd_cli_context: chat template example:
USER: You are a helpful assistant
Hello
 ASSISTANT:Hi thereUSER: How are you?
 ASSISTANT:
0.10.994.531 W init_audio: audio input is in experimental stage and may have reduced quality:
    https://github.com/ggml-org/llama.cpp/discussions/13759
0.10.994.776 I main: loading model: /Users/ghart/models/ibm-granite/granite-speech-4.1-2b-GGUF/granite-speech-4.1-2b-Q4_K_M.gguf
0.10.994.780 W WARN: This is an experimental CLI for testing multimodal capability.
0.10.994.780 W       For normal use cases, please use the standard llama-cli

総合的研究成果は、国際的な研究成果を発表するために必要です。

./build/bin/llama-mtmd-cli -m ~/models/ibm-granite/granite-speech-4.1-2b-GGUF/granite-speech-4.1-2b-Q4_K_M.gguf --mmproj ~/models/ibm-granite/granite-speech-4.1-2b-GGUF/mmproj-model-f16.gguf  --temp 0 -n 128 --flash-attn on --audio ~/models/ibm-granite/granite-speech-4.1-2b/multilingual_sample.wav -p "transcribe this audio to a text format" --jinja
0.00.033.746 I common_init_result: fitting params to device memory ...
0.00.033.748 I common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on)
0.01.091.898 I common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
0.01.124.713 I mtmd_cli_context: chat template example:
USER: You are a helpful assistant
Hello
 ASSISTANT:Hi thereUSER: How are you?
 ASSISTANT:
0.11.259.268 W init_audio: audio input is in experimental stage and may have reduced quality:
    https://github.com/ggml-org/llama.cpp/discussions/13759
0.11.259.528 I main: loading model: /Users/ghart/models/ibm-granite/granite-speech-4.1-2b-GGUF/granite-speech-4.1-2b-Q4_K_M.gguf
0.11.259.535 W WARN: This is an experimental CLI for testing multimodal capability.
0.11.259.535 W       For normal use cases, please use the standard llama-cli

for timothy was a spoiled cat, and he allowed no one to interfere. everybody waited upon him, moving their chairs even, for he was monarch of the hearth. "dinarzade, la nuit suivante appela sa soeur quand il en fut temps. 'si vous ne dormez pas, ma soeur,' lui dit elle, 'je vous prie en attendant le jour qui paraîtra bientôt de continuer le conte du pêcheur.