Unable to get rid of chat

[spamshaker]

lama-cli --vision-gemma-12b-default -p describe --image img.png > desc.md

falls into the chat without passing stream to desc.md

[taronaeo]

without passing stream to desc.md

I recall seeing this not being implemented in code somewhere after llama-cli was refactored. But can you try adding this before the pipe?

2>&1

Not tested btw.

[CISC]

The new llama-cli does not support this, I think @ngxson suggested implementing --output-file instead.

[spamshaker]

2>&1

This will just redirect spam to stdout, that I would want to avoid :)

The new llama-cli does not support this, I think @ngxson suggested implementing --output-file instead.

--output-file totally make sense I would also say -o for short would be perfect :)

also there is the flag -st that runs and close as one shot.

llama.cpp/tools/cli/README.md

Line 154 in 382808c

| `-st, --single-turn` | run conversation for a single turn only, then exit when done<br/>will not be interactive if first turn is predefined with --prompt<br/>(default: false) |

maybe this would work if the there is -st and -p together then don't attach stdin

[CISC]

The new llama-cli does not support this, I think @ngxson suggested implementing --output-file instead.

--output-file totally make sense I would also say -o for short would be perfect :)

Yes, that is the short form:

llama.cpp/common/arg.cpp

Lines 2472 to 2478 in 0bd1212

	add_opt(common_arg(
	{"-o", "--output", "--output-file"}, "FNAME",
	string_format("output file (default: '%s')", params.out_file.c_str()),
	[](common_params & params, const std::string & value) {
	params.out_file = value;
	}
	).set_examples({LLAMA_EXAMPLE_IMATRIX, LLAMA_EXAMPLE_CVECTOR_GENERATOR, LLAMA_EXAMPLE_EXPORT_LORA, LLAMA_EXAMPLE_TTS, LLAMA_EXAMPLE_FINETUNE}));