UPSTREAM PR #18954: model-conversion : add tensor-info.py utility

[loci-dev]

Mirrored from ggml-org/llama.cpp#18954

This commit adds a new python script that can be used to print tensors information from a tensor in a safetensors model.

The motivation for this is that during model conversion work it can sometimes be useful to verify the shape of tensors in the original model. While it is possible to print the tensors when loading the model this can be slow when working with larger models.
With this script it is possible to quickly query tensor shapes.

Example usage:

(venv) $ ./scripts/utils/tensor-info.py --help
usage: tensor-info.py [-h] [-m MODEL_PATH] [-l] [tensor_name]

Print tensor information from a safetensors model

positional arguments:
  tensor_name           Name of the tensor to inspect

options:
  -h, --help            show this help message and exit
  -m MODEL_PATH, --model-path MODEL_PATH
                        Path to the model directory (default: MODEL_PATH environment variable)
  -l, --list            List unique tensor patterns in the model (layer numbers replaced with #)

Listing tensor names:

(venv) $ ./scripts/utils/tensor-info.py -m ~/work/ai/models/google/embeddinggemma-300m -l
embed_tokens.weight
layers.#.input_layernorm.weight
layers.#.mlp.down_proj.weight
layers.#.mlp.gate_proj.weight
layers.#.mlp.up_proj.weight
layers.#.post_attention_layernorm.weight
layers.#.post_feedforward_layernorm.weight
layers.#.pre_feedforward_layernorm.weight
layers.#.self_attn.k_norm.weight
layers.#.self_attn.k_proj.weight
layers.#.self_attn.o_proj.weight
layers.#.self_attn.q_norm.weight
layers.#.self_attn.q_proj.weight
layers.#.self_attn.v_proj.weight
norm.weight

Printing a specific tensor's information:

(venv) $ ./scripts/utils/tensor-info.py -m ~/work/ai/models/google/embeddinggemma-300m layers.0.input_layernorm.weight
Tensor: layers.0.input_layernorm.weight
File:   model.safetensors
Shape:  [768]

[loci-review]

Explore the complete analysis inside the Version Insights

Based on the analysis, no functions were identified with meaningful performance changes between the base and target versions. The code modifications did not result in measurable performance impacts to response time or throughput time metrics.