GGUF Parser

tl;dr, Review/Check GGUF files and estimate the memory usage.

GGUF is a file format for storing models for inference with GGML and executors based on GGML. GGUF is a binary format that is designed for fast loading and saving of models, and for ease of reading. Models are traditionally developed using PyTorch or another framework, and then converted to GGUF for use in GGML.

GGUF Parser helps in reviewing and estimating the usage and maximum tokens per second of a GGUF format model without download it.

Key Features

No File Required: GGUF Parser uses chunking reading to parse the metadata of remote GGUF file, which means you don't need to download the entire file and load it.
Accurate Prediction: The evaluation results of GGUF Parser usually deviate from the actual usage by about 100MiB.
Quick Verification: Provide device metrics to calculate the maximum tokens per second (TPS) without running the model.
Fast: GGUF Parser is written in Go, which is fast and efficient.

Notes

GGUF Parser estimates the maximum tokens per second(MAX TPS) for a model (experimental).
GGUF Parser distinguishes the remote devices from --tensor-split via --rpc.
- For one host multiple GPU devices, you can use --tensor-split to get the estimated memory usage of each GPU.
- For multiple hosts multiple GPU devices, you can use --tensor-split and --rpc to get the estimated memory usage of each GPU. Since v0.11.0, --rpc flag masks the devices specified by --tensor-split in front.
Table result usage:
- I/T/O indicates the count for input layers, transformer layers, and output layers. Input layers are not offloaded at present.
- DISTRIBUTABLE indicates the GGUF file supports distribution inference or not, if the file doesn't support distribution inference, you can not offload it with RPC servers.
- RAM indicates the system memory usage when running LLaMA.Cpp or LLaMA.Cpp like application.
- VRAM * indicates the local GPU memory usage when serving the GGUF file.
- RPC * (V)RAM indicates the remote GPU memory usage when serving the GGUF file.
- UMA indicates the memory usage of Apple macOS only. NONUMA adapts to other cases, including none GPU devices.

Installation

Install from releases.

Overview

Parse

Parse Local File

$ gguf-parser --path="~/.cache/lm-studio/models/NousResearch/Hermes-2-Pro-Mistral-7B-GGUF/Hermes-2-Pro-Mistral-7B.Q5_K_M.gguf"
+-------------------------------------------------------------------------------------------+
| METADATA                                                                                  |
+-------+-------+-------+----------------+---------------+----------+------------+----------+
|  TYPE |  NAME |  ARCH |  QUANTIZATION  | LITTLE ENDIAN |   SIZE   | PARAMETERS |    BPW   |
+-------+-------+-------+----------------+---------------+----------+------------+----------+
| model | jeffq | llama | IQ3_XXS/Q5_K_M |      true     | 4.78 GiB |   7.24 B   | 5.67 bpw |
+-------+-------+-------+----------------+---------------+----------+------------+----------+

+---------------------------------------------------------------------------------------------------------------------------------------------------+
| ARCHITECTURE                                                                                                                                      |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
| MAX CONTEXT LEN | EMBEDDING LEN | EMBEDDING GQA | ATTENTION CAUSAL | ATTENTION HEAD CNT | LAYERS | FEED FORWARD LEN | EXPERT CNT | VOCABULARY LEN |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
|      32768      |      4096     |       4       |       true       |         32         |   32   |       14336      |      0     |      32032     |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+

+-------------------------------------------------------------------------------------------------------------------------------------------------------+
| TOKENIZER                                                                                                                                             |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| MODEL | TOKENS SIZE | TOKENS LEN | ADDED TOKENS LEN | BOS TOKEN | EOS TOKEN | EOT TOKEN | EOM TOKEN | UNKNOWN TOKEN | SEPARATOR TOKEN | PADDING TOKEN |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| llama |  450.50 KiB |    32032   |        N/A       |     1     |   32000   |    N/A    |    N/A    |      N/A      |       N/A       |      N/A      |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+

+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                        |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+----------------------------------------------+-------------------------------------+
|  ARCH | CONTEXT SIZE | BATCH SIZE (L / P) | FLASH ATTENTION | MMAP LOAD | EMBEDDING ONLY | DISTRIBUTABLE | OFFLOAD LAYERS | FULL OFFLOADED |                      RAM                     |                VRAM 0               |
|       |              |                    |                 |           |                |               |                |                +--------------------+------------+------------+----------------+--------+-----------+
|       |              |                    |                 |           |                |               |                |                | LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA  |   NONUMA  |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+--------+-----------+
| llama |     32768    |     2048 / 512     |     Disabled    |  Enabled  |       No       |   Supported   |   33 (32 + 1)  |       Yes      |      1 + 0 + 0     | 168.25 MiB | 318.25 MiB |     32 + 1     |  4 GiB | 11.16 GiB |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+--------+-----------+

$ # Retrieve the model's metadata via split file,
$ # which needs all split files has been downloaded.
$ gguf-parser --path="~/.cache/lm-studio/models/Qwen/Qwen2-72B-Instruct-GGUF/qwen2-72b-instruct-q6_k-00001-of-00002.gguf"

+------------------------------------------------------------------------------------------------------------+
| METADATA                                                                                                   |
+-------+-------------------------+-------+--------------+---------------+-----------+------------+----------+
|  TYPE |           NAME          |  ARCH | QUANTIZATION | LITTLE ENDIAN |    SIZE   | PARAMETERS |    BPW   |
+-------+-------------------------+-------+--------------+---------------+-----------+------------+----------+
| model | 72b.5000B--cmix31-ba... | qwen2 |  IQ1_S/Q6_K  |      true     | 59.92 GiB |   72.71 B  | 7.08 bpw |
+-------+-------------------------+-------+--------------+---------------+-----------+------------+----------+

+---------------------------------------------------------------------------------------------------------------------------------------------------+
| ARCHITECTURE                                                                                                                                      |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
| MAX CONTEXT LEN | EMBEDDING LEN | EMBEDDING GQA | ATTENTION CAUSAL | ATTENTION HEAD CNT | LAYERS | FEED FORWARD LEN | EXPERT CNT | VOCABULARY LEN |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
|      32768      |      8192     |       8       |       true       |         64         |   80   |       29568      |      0     |     152064     |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+

+-------------------------------------------------------------------------------------------------------------------------------------------------------+
| TOKENIZER                                                                                                                                             |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| MODEL | TOKENS SIZE | TOKENS LEN | ADDED TOKENS LEN | BOS TOKEN | EOS TOKEN | EOT TOKEN | EOM TOKEN | UNKNOWN TOKEN | SEPARATOR TOKEN | PADDING TOKEN |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
|  gpt2 |   2.47 MiB  |   152064   |        N/A       |   151643  |   151645  |    N/A    |    N/A    |      N/A      |       N/A       |     151643    |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+

+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                        |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+----------------------------------------------+-------------------------------------+
|  ARCH | CONTEXT SIZE | BATCH SIZE (L / P) | FLASH ATTENTION | MMAP LOAD | EMBEDDING ONLY | DISTRIBUTABLE | OFFLOAD LAYERS | FULL OFFLOADED |                      RAM                     |                VRAM 0               |
|       |              |                    |                 |           |                |               |                |                +--------------------+------------+------------+----------------+--------+-----------+
|       |              |                    |                 |           |                |               |                |                | LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA  |   NONUMA  |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+--------+-----------+
| qwen2 |     32768    |     2048 / 512     |     Disabled    |  Enabled  |       No       |  Unsupported  |   81 (80 + 1)  |       Yes      |      1 + 0 + 0     | 291.38 MiB | 441.38 MiB |     80 + 1     | 10 GiB | 73.47 GiB |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+--------+-----------+

Parse Remote File

$ gguf-parser --url="https://huggingface.co/NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO-GGUF/resolve/main/Nous-Hermes-2-Mixtral-8x7B-DPO.Q3_K_M.gguf"
+------------------------------------------------------------------------------------------+
| METADATA                                                                                 |
+-------+----------+-------+--------------+---------------+--------+------------+----------+
|  TYPE |   NAME   |  ARCH | QUANTIZATION | LITTLE ENDIAN |  SIZE  | PARAMETERS |    BPW   |
+-------+----------+-------+--------------+---------------+--------+------------+----------+
| model | emozilla | llama |  Q4_K/Q3_K_M |      true     | 21 GiB |   46.70 B  | 3.86 bpw |
+-------+----------+-------+--------------+---------------+--------+------------+----------+

+---------------------------------------------------------------------------------------------------------------------------------------------------+
| ARCHITECTURE                                                                                                                                      |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
| MAX CONTEXT LEN | EMBEDDING LEN | EMBEDDING GQA | ATTENTION CAUSAL | ATTENTION HEAD CNT | LAYERS | FEED FORWARD LEN | EXPERT CNT | VOCABULARY LEN |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
|      32768      |      4096     |       4       |       true       |         32         |   32   |       14336      |      8     |      32002     |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+

+-------------------------------------------------------------------------------------------------------------------------------------------------------+
| TOKENIZER                                                                                                                                             |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| MODEL | TOKENS SIZE | TOKENS LEN | ADDED TOKENS LEN | BOS TOKEN | EOS TOKEN | EOT TOKEN | EOM TOKEN | UNKNOWN TOKEN | SEPARATOR TOKEN | PADDING TOKEN |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| llama |  449.91 KiB |    32002   |        N/A       |     1     |   32000   |    N/A    |    N/A    |       0       |       N/A       |       2       |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+

+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                             |
+-------+--------------+--------------------+-----------------+-------------+----------------+---------------+----------------+----------------+----------------------------------------------+----------------------------------------+
|  ARCH | CONTEXT SIZE | BATCH SIZE (L / P) | FLASH ATTENTION |  MMAP LOAD  | EMBEDDING ONLY | DISTRIBUTABLE | OFFLOAD LAYERS | FULL OFFLOADED |                      RAM                     |                 VRAM 0                 |
|       |              |                    |                 |             |                |               |                |                +--------------------+------------+------------+----------------+-----------+-----------+
|       |              |                    |                 |             |                |               |                |                | LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |    UMA    |   NONUMA  |
+-------+--------------+--------------------+-----------------+-------------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+-----------+-----------+
| llama |     32768    |     2048 / 512     |     Disabled    | Unsupported |       No       |   Supported   |   33 (32 + 1)  |       Yes      |      1 + 0 + 0     | 269.10 MiB | 419.10 MiB |     32 + 1     | 24.94 GiB | 27.41 GiB |
+-------+--------------+--------------------+-----------------+-------------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+-----------+-----------+

$ # Retrieve the model's metadata via split file

$ gguf-parser --url="https://huggingface.co/MaziyarPanahi/Meta-Llama-3.1-405B-Instruct-GGUF/resolve/main/Meta-Llama-3.1-405B-Instruct.Q2_K.gguf-00001-of-00009.gguf"
+-------------------------------------------------------------------------------------------------------------+
| METADATA                                                                                                    |
+-------+-------------------------+-------+--------------+---------------+------------+------------+----------+
|  TYPE |           NAME          |  ARCH | QUANTIZATION | LITTLE ENDIAN |    SIZE    | PARAMETERS |    BPW   |
+-------+-------------------------+-------+--------------+---------------+------------+------------+----------+
| model | Models Meta Llama Me... | llama |     Q2_K     |      true     | 140.81 GiB |  410.08 B  | 2.95 bpw |
+-------+-------------------------+-------+--------------+---------------+------------+------------+----------+

+---------------------------------------------------------------------------------------------------------------------------------------------------+
| ARCHITECTURE                                                                                                                                      |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
| MAX CONTEXT LEN | EMBEDDING LEN | EMBEDDING GQA | ATTENTION CAUSAL | ATTENTION HEAD CNT | LAYERS | FEED FORWARD LEN | EXPERT CNT | VOCABULARY LEN |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
|      131072     |     16384     |       8       |       true       |         128        |   126  |       53248      |      0     |     128256     |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+

+-------------------------------------------------------------------------------------------------------------------------------------------------------+
| TOKENIZER                                                                                                                                             |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| MODEL | TOKENS SIZE | TOKENS LEN | ADDED TOKENS LEN | BOS TOKEN | EOS TOKEN | EOT TOKEN | EOM TOKEN | UNKNOWN TOKEN | SEPARATOR TOKEN | PADDING TOKEN |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
|  gpt2 |    2 MiB    |   128256   |        N/A       |   128000  |   128009  |    N/A    |    N/A    |      N/A      |       N/A       |      N/A      |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+

+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                          |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+----------------------------------------------+---------------------------------------+
|  ARCH | CONTEXT SIZE | BATCH SIZE (L / P) | FLASH ATTENTION | MMAP LOAD | EMBEDDING ONLY | DISTRIBUTABLE | OFFLOAD LAYERS | FULL OFFLOADED |                      RAM                     |                 VRAM 0                |
|       |              |                    |                 |           |                |               |                |                +--------------------+------------+------------+----------------+---------+------------+
|       |              |                    |                 |           |                |               |                |                | LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA   |   NONUMA   |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+---------+------------+
| llama |    131072    |     2048 / 512     |     Disabled    |  Enabled  |       No       |   Supported   |  127 (126 + 1) |       Yes      |      1 + 0 + 0     | 652.53 MiB | 802.53 MiB |     126 + 1    | 126 GiB | 299.79 GiB |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+---------+------------+

Parse From HuggingFace

$ gguf-parser --hf-repo="openbmb/MiniCPM-Llama3-V-2_5-gguf" --hf-file="ggml-model-Q5_K_M.gguf" --hf-mmproj-file="mmproj-model-f16.gguf"
+-------------------------------------------------------------------------------------------+
| METADATA                                                                                  |
+-------+-------+-------+----------------+---------------+----------+------------+----------+
|  TYPE |  NAME |  ARCH |  QUANTIZATION  | LITTLE ENDIAN |   SIZE   | PARAMETERS |    BPW   |
+-------+-------+-------+----------------+---------------+----------+------------+----------+
| model | model | llama | IQ3_XXS/Q5_K_M |      true     | 5.33 GiB |   8.03 B   | 5.70 bpw |
+-------+-------+-------+----------------+---------------+----------+------------+----------+

+---------------------------------------------------------------------------------------------------------------------------------------------------+
| ARCHITECTURE                                                                                                                                      |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
| MAX CONTEXT LEN | EMBEDDING LEN | EMBEDDING GQA | ATTENTION CAUSAL | ATTENTION HEAD CNT | LAYERS | FEED FORWARD LEN | EXPERT CNT | VOCABULARY LEN |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
|       8192      |      4096     |       4       |       true       |         32         |   32   |       14336      |      0     |     128256     |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+

+-------------------------------------------------------------------------------------------------------------------------------------------------------+
| TOKENIZER                                                                                                                                             |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| MODEL | TOKENS SIZE | TOKENS LEN | ADDED TOKENS LEN | BOS TOKEN | EOS TOKEN | EOT TOKEN | EOM TOKEN | UNKNOWN TOKEN | SEPARATOR TOKEN | PADDING TOKEN |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
|  gpt2 |    2 MiB    |   128256   |        N/A       |   128000  |   128001  |    N/A    |    N/A    |     128002    |       N/A       |       0       |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+

+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                       |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+----------------------------------------------+------------------------------------+
|  ARCH | CONTEXT SIZE | BATCH SIZE (L / P) | FLASH ATTENTION | MMAP LOAD | EMBEDDING ONLY | DISTRIBUTABLE | OFFLOAD LAYERS | FULL OFFLOADED |                      RAM                     |               VRAM 0               |
|       |              |                    |                 |           |                |               |                |                +--------------------+------------+------------+----------------+--------+----------+
|       |              |                    |                 |           |                |               |                |                | LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA  |  NONUMA  |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+--------+----------+
| llama |     8192     |     2048 / 512     |     Disabled    |  Enabled  |       No       |   Supported   |   33 (32 + 1)  |       Yes      |      1 + 0 + 0     | 176.85 MiB | 326.85 MiB |     32 + 1     |  1 GiB | 7.78 GiB |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+--------+----------+

$ # Retrieve the model's metadata via split file

$ gguf-parser --hf-repo="etemiz/Llama-3.1-405B-Inst-GGUF" --hf-file="llama-3.1-405b-IQ1_M-00019-of-00019.gguf"
+------------------------------------------------------------------------------------------------------------+
| METADATA                                                                                                   |
+-------+-------------------------+-------+--------------+---------------+-----------+------------+----------+
|  TYPE |           NAME          |  ARCH | QUANTIZATION | LITTLE ENDIAN |    SIZE   | PARAMETERS |    BPW   |
+-------+-------------------------+-------+--------------+---------------+-----------+------------+----------+
| model | Meta-Llama-3.1-405B-... | llama |     IQ1_M    |      true     | 88.61 GiB |  410.08 B  | 1.86 bpw |
+-------+-------------------------+-------+--------------+---------------+-----------+------------+----------+

+---------------------------------------------------------------------------------------------------------------------------------------------------+
| ARCHITECTURE                                                                                                                                      |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
| MAX CONTEXT LEN | EMBEDDING LEN | EMBEDDING GQA | ATTENTION CAUSAL | ATTENTION HEAD CNT | LAYERS | FEED FORWARD LEN | EXPERT CNT | VOCABULARY LEN |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
|      131072     |     16384     |       8       |       true       |         128        |   126  |       53248      |      0     |     128256     |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+

+-------------------------------------------------------------------------------------------------------------------------------------------------------+
| TOKENIZER                                                                                                                                             |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| MODEL | TOKENS SIZE | TOKENS LEN | ADDED TOKENS LEN | BOS TOKEN | EOS TOKEN | EOT TOKEN | EOM TOKEN | UNKNOWN TOKEN | SEPARATOR TOKEN | PADDING TOKEN |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
|  gpt2 |    2 MiB    |   128256   |        N/A       |   128000  |   128009  |    N/A    |    N/A    |      N/A      |       N/A       |      N/A      |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+

+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                          |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+----------------------------------------------+---------------------------------------+
|  ARCH | CONTEXT SIZE | BATCH SIZE (L / P) | FLASH ATTENTION | MMAP LOAD | EMBEDDING ONLY | DISTRIBUTABLE | OFFLOAD LAYERS | FULL OFFLOADED |                      RAM                     |                 VRAM 0                |
|       |              |                    |                 |           |                |               |                |                +--------------------+------------+------------+----------------+---------+------------+
|       |              |                    |                 |           |                |               |                |                | LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA   |   NONUMA   |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+---------+------------+
| llama |    131072    |     2048 / 512     |     Disabled    |  Enabled  |       No       |   Supported   |  127 (126 + 1) |       Yes      |      1 + 0 + 0     | 652.53 MiB | 802.53 MiB |     126 + 1    | 126 GiB | 247.59 GiB |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+---------+------------+

Parse From ModelScope

$ gguf-parser --ms-repo="shaowenchen/chinese-alpaca-2-13b-16k-gguf" --ms-file="chinese-alpaca-2-13b-16k.Q5_K.gguf"
+------------------------------------------------------------------------------------------+
| METADATA                                                                                 |
+-------+------+-------+----------------+---------------+----------+------------+----------+
|  TYPE | NAME |  ARCH |  QUANTIZATION  | LITTLE ENDIAN |   SIZE   | PARAMETERS |    BPW   |
+-------+------+-------+----------------+---------------+----------+------------+----------+
| model |  ..  | llama | IQ3_XXS/Q5_K_M |      true     | 8.76 GiB |   13.25 B  | 5.68 bpw |
+-------+------+-------+----------------+---------------+----------+------------+----------+

+---------------------------------------------------------------------------------------------------------------------------------------------------+
| ARCHITECTURE                                                                                                                                      |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
| MAX CONTEXT LEN | EMBEDDING LEN | EMBEDDING GQA | ATTENTION CAUSAL | ATTENTION HEAD CNT | LAYERS | FEED FORWARD LEN | EXPERT CNT | VOCABULARY LEN |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
|      16384      |      5120     |       1       |       true       |         N/A        |   40   |       13824      |      0     |      55296     |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+

+-------------------------------------------------------------------------------------------------------------------------------------------------------+
| TOKENIZER                                                                                                                                             |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| MODEL | TOKENS SIZE | TOKENS LEN | ADDED TOKENS LEN | BOS TOKEN | EOS TOKEN | EOT TOKEN | EOM TOKEN | UNKNOWN TOKEN | SEPARATOR TOKEN | PADDING TOKEN |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| llama |  769.83 KiB |    55296   |        N/A       |     1     |     2     |    N/A    |    N/A    |      N/A      |       N/A       |      N/A      |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+

+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                           |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+----------------------------------------------+----------------------------------------+
|  ARCH | CONTEXT SIZE | BATCH SIZE (L / P) | FLASH ATTENTION | MMAP LOAD | EMBEDDING ONLY | DISTRIBUTABLE | OFFLOAD LAYERS | FULL OFFLOADED |                      RAM                     |                 VRAM 0                 |
|       |              |                    |                 |           |                |               |                |                +--------------------+------------+------------+----------------+-----------+-----------+
|       |              |                    |                 |           |                |               |                |                | LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |    UMA    |   NONUMA  |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+-----------+-----------+
| llama |     16384    |     2048 / 512     |     Disabled    |  Enabled  |       No       |   Supported   |   41 (40 + 1)  |       Yes      |      1 + 0 + 0     | 144.95 MiB | 294.95 MiB |     40 + 1     | 12.50 GiB | 22.96 GiB |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+-----------+-----------+

Parse From Ollama Library

$ gguf-parser --ol-model="llama3.1"
+-----------------------------------------------------------------------------------------------------------+
| METADATA                                                                                                  |
+-------+-------------------------+-------+--------------+---------------+----------+------------+----------+
|  TYPE |           NAME          |  ARCH | QUANTIZATION | LITTLE ENDIAN |   SIZE   | PARAMETERS |    BPW   |
+-------+-------------------------+-------+--------------+---------------+----------+------------+----------+
| model | Meta Llama 3.1 8B In... | llama |     Q4_0     |      true     | 4.33 GiB |   8.03 B   | 4.64 bpw |
+-------+-------------------------+-------+--------------+---------------+----------+------------+----------+

+---------------------------------------------------------------------------------------------------------------------------------------------------+
| ARCHITECTURE                                                                                                                                      |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
| MAX CONTEXT LEN | EMBEDDING LEN | EMBEDDING GQA | ATTENTION CAUSAL | ATTENTION HEAD CNT | LAYERS | FEED FORWARD LEN | EXPERT CNT | VOCABULARY LEN |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
|      131072     |      4096     |       4       |       true       |         32         |   32   |       14336      |      0     |     128256     |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+

+-------------------------------------------------------------------------------------------------------------------------------------------------------+
| TOKENIZER                                                                                                                                             |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| MODEL | TOKENS SIZE | TOKENS LEN | ADDED TOKENS LEN | BOS TOKEN | EOS TOKEN | EOT TOKEN | EOM TOKEN | UNKNOWN TOKEN | SEPARATOR TOKEN | PADDING TOKEN |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
|  gpt2 |    2 MiB    |   128256   |        N/A       |   128000  |   128009  |    N/A    |    N/A    |      N/A      |       N/A       |      N/A      |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+

+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                        |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+----------------------------------------------+-------------------------------------+
|  ARCH | CONTEXT SIZE | BATCH SIZE (L / P) | FLASH ATTENTION | MMAP LOAD | EMBEDDING ONLY | DISTRIBUTABLE | OFFLOAD LAYERS | FULL OFFLOADED |                      RAM                     |                VRAM 0               |
|       |              |                    |                 |           |                |               |                |                +--------------------+------------+------------+----------------+--------+-----------+
|       |              |                    |                 |           |                |               |                |                | LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA  |   NONUMA  |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+--------+-----------+
| llama |    131072    |     2048 / 512     |     Disabled    |  Enabled  |       No       |   Supported   |   33 (32 + 1)  |       Yes      |      1 + 0 + 0     | 403.62 MiB | 553.62 MiB |     32 + 1     | 16 GiB | 29.08 GiB |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+--------+-----------+

$ # Ollama Model includes the preset params and other artifacts, like multimodal projectors or LoRA adapters, 
$ # you can get the usage of Ollama running by using `--ol-usage` option.

$ gguf-parser --ol-model="llama3.1" --ol-usage
+-----------------------------------------------------------------------------------------------------------+
| METADATA                                                                                                  |
+-------+-------------------------+-------+--------------+---------------+----------+------------+----------+
|  TYPE |           NAME          |  ARCH | QUANTIZATION | LITTLE ENDIAN |   SIZE   | PARAMETERS |    BPW   |
+-------+-------------------------+-------+--------------+---------------+----------+------------+----------+
| model | Meta Llama 3.1 8B In... | llama |     Q4_0     |      true     | 4.33 GiB |   8.03 B   | 4.64 bpw |
+-------+-------------------------+-------+--------------+---------------+----------+------------+----------+

+---------------------------------------------------------------------------------------------------------------------------------------------------+
| ARCHITECTURE                                                                                                                                      |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
| MAX CONTEXT LEN | EMBEDDING LEN | EMBEDDING GQA | ATTENTION CAUSAL | ATTENTION HEAD CNT | LAYERS | FEED FORWARD LEN | EXPERT CNT | VOCABULARY LEN |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+
|      131072     |      4096     |       4       |       true       |         32         |   32   |       14336      |      0     |     128256     |
+-----------------+---------------+---------------+------------------+--------------------+--------+------------------+------------+----------------+

+-------------------------------------------------------------------------------------------------------------------------------------------------------+
| TOKENIZER                                                                                                                                             |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
| MODEL | TOKENS SIZE | TOKENS LEN | ADDED TOKENS LEN | BOS TOKEN | EOS TOKEN | EOT TOKEN | EOM TOKEN | UNKNOWN TOKEN | SEPARATOR TOKEN | PADDING TOKEN |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+
|  gpt2 |    2 MiB    |   128256   |        N/A       |   128000  |   128009  |    N/A    |    N/A    |      N/A      |       N/A       |      N/A      |
+-------+-------------+------------+------------------+-----------+-----------+-----------+-----------+---------------+-----------------+---------------+

+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                           |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+----------------------------------------------+----------------------------------------+
|  ARCH | CONTEXT SIZE | BATCH SIZE (L / P) | FLASH ATTENTION | MMAP LOAD | EMBEDDING ONLY | DISTRIBUTABLE | OFFLOAD LAYERS | FULL OFFLOADED |                      RAM                     |                 VRAM 0                 |
|       |              |                    |                 |           |                |               |                |                +--------------------+------------+------------+----------------+------------+----------+
|       |              |                    |                 |           |                |               |                |                | LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |     UMA    |  NONUMA  |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+------------+----------+
| llama |     2048     |     2048 / 512     |     Disabled    |  Enabled  |       No       |   Supported   |   33 (32 + 1)  |       Yes      |      1 + 0 + 0     | 151.62 MiB | 301.62 MiB |     32 + 1     | 256.50 MiB | 4.82 GiB |
+-------+--------------+--------------------+-----------------+-----------+----------------+---------------+----------------+----------------+--------------------+------------+------------+----------------+------------+----------+

Parse None Model

$ # Parse Multi-Modal Projector
$ gguf-parser --hf-repo="xtuner/llava-llama-3-8b-v1_1-gguf" --hf-file="llava-llama-3-8b-v1_1-mmproj-f16.gguf"                                                                        
+-----------------------------------------------------------------------------------------------------------------+
| METADATA                                                                                                        |
+-----------+-------------------------+------+--------------+---------------+------------+------------+-----------+
|    TYPE   |           NAME          | ARCH | QUANTIZATION | LITTLE ENDIAN |    SIZE    | PARAMETERS |    BPW    |
+-----------+-------------------------+------+--------------+---------------+------------+------------+-----------+
| projector | openai/clip-vit-larg... | clip |      F16     |      true     | 595.49 MiB |  311.89 M  | 16.02 bpw |
+-----------+-------------------------+------+--------------+---------------+------------+------------+-----------+

+----------------------------------------------------------------------+
| ARCHITECTURE                                                         |
+----------------+---------------+--------+------------------+---------+
| PROJECTOR TYPE | EMBEDDING LEN | LAYERS | FEED FORWARD LEN | ENCODER |
+----------------+---------------+--------+------------------+---------+
|       mlp      |      1024     |   23   |       4096       |  Vision |
+----------------+---------------+--------+------------------+---------+

$ # Parse LoRA Adapter
$ gguf-parser --hf-repo="ngxson/test_gguf_lora_adapter" --hf-file="lora-Llama-3-Instruct-abliteration-LoRA-8B-f16.gguf"
+---------------------------------------------------------------------------------------------+
| METADATA                                                                                    |
+---------+------+-------+--------------+---------------+------------+------------+-----------+
|   TYPE  | NAME |  ARCH | QUANTIZATION | LITTLE ENDIAN |    SIZE    | PARAMETERS |    BPW    |
+---------+------+-------+--------------+---------------+------------+------------+-----------+
| adapter |  N/A | llama |      F16     |      true     | 168.08 MiB |   88.12 M  | 16.00 bpw |
+---------+------+-------+--------------+---------------+------------+------------+-----------+

+---------------------------+
| ARCHITECTURE              |
+--------------+------------+
| ADAPTER TYPE | LORA ALPHA |
+--------------+------------+
|     lora     |     32     |
+--------------+------------+

Estimate

Across Multiple GPU Devices

Imaging you're preparing to run the hierholzer/Llama-3.1-70B-Instruct-GGUF model file across several hosts in your local network. Some of these hosts are equipped with GPU devices, while others do not have any GPU capabilities.

flowchart TD
    subgraph host4["Windows 11 (host4)"]
        ram40(["11GiB RAM remaining"])
    end
    subgraph host3["Apple macOS (host3)"]
        gpu10["Apple M1 Max (6GiB VRAM remaining)"]
    end
    subgraph host2["Windows 11 (host2)"]
        gpu20["NVIDIA 4090 (12GiB VRAM remaining)"]
    end
    subgraph host1["Ubuntu (host1)"]
        gpu30["NVIDIA 4080 0 (8GiB VRAM remaining)"]
        gpu31["NVIDIA 4080 1 (10GiB VRAM remaining)"]
    end

Single Host Multiple GPU Devices

Let's assume you plan to run the model on host1 only.

flowchart TD
    subgraph host1["Ubuntu (host1)"]
        gpu30["NVIDIA 4080 0 (8GiB VRAM remaining)"]
        gpu31["NVIDIA 4080 1 (10GiB VRAM remaining)"]
    end

$ gguf-parser --hf-repo="hierholzer/Llama-3.1-70B-Instruct-GGUF" --hf-file="Llama-3.1-70B-Instruct-Q4_K_M.gguf" --skip-metadata --skip-architecture --skip-tokenizer --ctx-size=1024 --tensor-split="8,10" --in-short
+------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                     |
+----------------------------------------------+--------------------------------------+----------------------------------------+
|                      RAM                     |                VRAM 0                |                 VRAM 1                 |
+--------------------+------------+------------+----------------+---------+-----------+----------------+-----------+-----------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA   |   NONUMA  | LAYERS (T + O) |    UMA    |   NONUMA  |
+--------------------+------------+------------+----------------+---------+-----------+----------------+-----------+-----------+
|      1 + 0 + 0     | 238.08 MiB | 388.08 MiB |     36 + 0     | 144 MiB | 17.79 GiB |     44 + 1     | 22.01 GiB | 22.51 GiB |
+--------------------+------------+------------+----------------+---------+-----------+----------------+-----------+-----------+

Based on the output provided, serving the hierholzer/Llama-3.1-70B-Instruct-GGUF model on host1 has the following resource consumption:

Host	Available RAM	Request RAM	Available VRAM	Request VRAM	Result
host1	ENOUGH	388.08 MiB			👍
host1 (NVIDIA 4080 0)			8 GiB	17.79 GiB
host1 (NVIDIA 4080 1)			10 GiB	22.51 GiB

It appears that running the model on host1 alone is not feasible.

Multiple Hosts Multiple GPU Devices

Next, let's consider the scenario where you plan to run the model on host4, while offloading all layers to host1, host2, and host3.

flowchart TD
    host4 -->|TCP| gpu10
    host4 -->|TCP| gpu20
    host4 -->|TCP| gpu30
    host4 -->|TCP| gpu31

    subgraph host4["Windows 11 (host4)"]
        ram40(["11GiB RAM remaining"])
    end
    subgraph host3["Apple macOS (host3)"]
        gpu10["Apple M1 Max (6GiB VRAM remaining)"]
    end
    subgraph host2["Windows 11 (host2)"]
        gpu20["NVIDIA 4090 (12GiB VRAM remaining)"]
    end
    subgraph host1["Ubuntu (host1)"]
        gpu30["NVIDIA 4080 0 (8GiB VRAM remaining)"]
        gpu31["NVIDIA 4080 1 (10GiB VRAM remaining)"]
    end

$ gguf-parser --hf-repo="hierholzer/Llama-3.1-70B-Instruct-GGUF" --hf-file="Llama-3.1-70B-Instruct-Q4_K_M.gguf" --skip-metadata --skip-architecture --skip-tokenizer --ctx-size=1024 --tensor-split="8,10,12,6" --rpc="host1:50052,host1:50053,host2:50052,host3:50052" --in-short
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                                 |
+----------------------------------------------+----------------------------------------------+----------------------------------------------+----------------------------------------------+----------------------------------------------+
|                      RAM                     |                 RPC 0 (V)RAM                 |                 RPC 1 (V)RAM                 |                 RPC 2 (V)RAM                 |                 RPC 3 (V)RAM                 |
+--------------------+------------+------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |      UMA     |    NONUMA    | LAYERS (T + O) |      UMA     |    NONUMA    | LAYERS (T + O) |      UMA     |    NONUMA    | LAYERS (T + O) |      UMA     |    NONUMA    |
+--------------------+------------+------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+
|      1 + 0 + 0     | 238.08 MiB | 388.08 MiB |     18 + 0     |   8.85 GiB   |   9.28 GiB   |     23 + 0     |   10.88 GiB  |   11.32 GiB  |     27 + 0     |   12.75 GiB  |   13.19 GiB  |     12 + 1     |   6.87 GiB   |   7.38 GiB   |
+--------------------+------------+------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+

According to the output provided, serving the hierholzer/Llama-3.1-70B-Instruct-GGUF model on host4 results in the following resource consumption:

Host	Available RAM	Request RAM	Available VRAM	Request VRAM	Result
host4	11 GiB	388.08 MiB			👍
host1 (NVIDIA 4080 0)			8 GiB	9.28 GiB
host1 (NVIDIA 4080 1)			10 GiB	11.32 GiB
host2 (NVIDIA 4090)			12 GiB	13.19 GiB
host3 (Apple M1 Max)	ENOUGH		6 GiB	6.87 GiB

It seems that the model cannot be served on host4, even with all layers offloaded to host1, host2, and host3.

We should consider a different approach: running the model on host3 while offloading all layers to host1, host2, and host4.

flowchart TD
    host3 -->|TCP| ram40
    host3 -->|TCP| gpu20
    host3 -->|TCP| gpu30
    host3 -->|TCP| gpu31

    subgraph host4["Windows 11 (host4)"]
        ram40(["11GiB RAM remaining"])
    end
    subgraph host3["Apple macOS (host3)"]
        gpu10["Apple M1 Max (6GiB VRAM remaining)"]
    end
    subgraph host2["Windows 11 (host2)"]
        gpu20["NVIDIA 4090 (12GiB VRAM remaining)"]
    end
    subgraph host1["Ubuntu (host1)"]
        gpu30["NVIDIA 4080 0 (8GiB VRAM remaining)"]
        gpu31["NVIDIA 4080 1 (10GiB VRAM remaining)"]
    end

$ gguf-parser --hf-repo="hierholzer/Llama-3.1-70B-Instruct-GGUF" --hf-file="Llama-3.1-70B-Instruct-Q4_K_M.gguf" --skip-metadata --skip-architecture --skip-tokenizer --ctx-size=1024 --tensor-split="11,12,8,10,6" --rpc="host4:50052,host2:50052,host1:50052,host1:50053" --in-short
+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| ESTIMATE                                                                                                                                                                                                                                                                         |
+----------------------------------------------+----------------------------------------------+----------------------------------------------+----------------------------------------------+----------------------------------------------+---------------------------------------+
|                      RAM                     |                 RPC 0 (V)RAM                 |                 RPC 1 (V)RAM                 |                 RPC 2 (V)RAM                 |                 RPC 3 (V)RAM                 |                 VRAM 0                |
+--------------------+------------+------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+-----------+----------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |      UMA     |    NONUMA    | LAYERS (T + O) |      UMA     |    NONUMA    | LAYERS (T + O) |      UMA     |    NONUMA    | LAYERS (T + O) |      UMA     |    NONUMA    | LAYERS (T + O) |    UMA    |  NONUMA  |
+--------------------+------------+------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+-----------+----------+
|      1 + 0 + 0     | 238.08 MiB | 388.08 MiB |     19 + 0     |   9.36 GiB   |   9.79 GiB   |     21 + 0     |   9.92 GiB   |   10.36 GiB  |     14 + 0     |   6.57 GiB   |   7.01 GiB   |     17 + 0     |   8.11 GiB   |   8.54 GiB   |      9 + 1     | 36.52 MiB | 5.91 GiB |
+--------------------+------------+------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+--------------+--------------+----------------+-----------+----------+

According to the output provided, serving the hierholzer/Llama-3.1-70B-Instruct-GGUF model on host3 results in the following resource consumption:

Host	Available RAM	Request RAM	Available VRAM	Request VRAM	Result
host3 (Apple M1 Max)	ENOUGH	238.08 MiB			👍
host4	11 GiB	9.79 GiB			👍
host2 (NVIDIA 4090)			12 GiB	10.36 GiB	👍
host1 (NVIDIA 4080 0)			8 GiB	7.01 GiB	👍
host1 (NVIDIA 4080 1)			10 GiB	8.54 GiB	👍
host3 (Apple M1 Max)			6 GiB	36.52 MiB	👍

Now, the model can be successfully served on host3, with all layers offloaded to host1, host2, and host4.

Maximum Tokens Per Second

The maximum TPS estimation for the GGUF Parser is determined by the model's parameter size, context size, model offloaded layers, and devices on which the model runs. Among these factors, the device's specifications are particularly important.

Inspired by LLM inference speed of light, GGUF Parser use the FLOPS and bandwidth of the device as evaluation metrics:

When the device is a CPU, FLOPS refers to the performance of that CPU, while bandwidth corresponds to the DRAM bandwidth.
When the device is a (i)GPU, FLOPS indicates the performance of that (i)GPU, and bandwidth corresponds to the VRAM bandwidth.
When the device is a specific host, FLOPS depends on whether the CPU or (i)GPU of that host is being used, while bandwidth corresponds to the bandwidth connecting the main node to that host. After all, a chain is only as strong as its weakest link. If the connection bandwidth between the main node and the host is equal to or greater than the *RAM bandwidth, then the bandwidth should be taken as the *RAM bandwidth value.

CPU FLOPS Calculation

The performance of a single CPU cache can be calculated using the following formula:

$$ CPU\ FLOPS = Number\ of \ Cores \times Core\ Frequency \times Floating\ Point\ Operations\ per\ Cycle $$

The Apple M1 Max CPU features a total of 10 cores, consisting of 8 performance cores and 2 efficiency cores. The performance cores operate at a clock speed of 3.2 GHz, while the efficiency cores run at 2.2 GHz. All cores support the ARM NEON instruction set, which enables 128-bit SIMD operations, allowing multiple floating-point numbers to be processed simultaneously within a single CPU cycle. Specifically, using single-precision (32-bit) floating-point numbers, each cycle can handle 4 floating-point operations.

The peak floating-point performance for a single performance core is calculated as follows:

$$ Peak\ Performance = 3.2\ GHz \times 4\ FLOPS = 12.8\ GFLOPS $$

For a single efficiency core, the calculation is:

$$ Peak\ Performance = 2.2\ GHz \times 4\ FLOPS = 8.8\ GFLOPS $$

Thus, the overall peak floating-point performance of the entire CPU can be determined by combining the contributions from both types of cores:

$$ Peak\ Performance = 8\ Cores \times 12.8\ GFLOPS + 2\ Cores \times 8.8\ GFLOPS = 120\ GFLOPS $$

This results in an average performance of 12 GFLOPS per core. It is evident that the average performance achieved by utilizing both performance and efficiency cores is lower than that obtained by exclusively using performance cores.

Run LLaMA2-7B-Chat with Apple Silicon M-series

Taking TheBloke/Llama-2-7B-Chat-GGUF as an example and estimate the maximum tokens per second for Apple Silicon M-series using the GGUF Parser.

$ # Estimate full offloaded Q8_0 model
$ gguf-parser --hf-repo TheBloke/LLaMA-7b-GGUF --hf-file llama-7b.Q8_0.gguf --skip-metadata --skip-architecture --skip-tokenizer --in-short \
  -c 512 \
  --device-metric "<CPU FLOPS>;<RAM BW>,<iGPU FLOPS>;<VRAM BW>"

$ # Estimate full offloaded Q4_0 model
$ gguf-parser --hf-repo TheBloke/LLaMA-7b-GGUF --hf-file llama-7b.Q4_0.gguf --skip-metadata --skip-architecture --skip-tokenizer --in-short \
  -c 512 \
  --device-metric "<CPU FLOPS>;<RAM BW>,<iGPU FLOPS>;<VRAM BW>"

Variant	CPU FLOPS (Performance Core)	iGPU FLOPS	(V)RAM Bandwidth	Q8_0 Max TPS	Q4_0 Max TPS
M1	51.2 GFLOPS (4 cores)	2.6 TFLOPS (8 cores)	68.3 GBps	8.68	14.56
M1 Pro	102.4 GFLOPS (8 cores)	5.2 TFLOPS (16 cores)	204.8 GBps	26.04	43.66
M1 Max	102.4 GFLOPS (8 cores)	10.4 TFLOPS (32 cores)	409.6 GBps	52.08	87.31
M1 Ultra	204.8 GFLOPS (16 cores)	21 TFLOPS (64 cores)	819.2 GBps	104.16	174.62
M2	56 GFLOPS (4 cores)	3.6 TFLOPS (10 cores)	102.4 GBps	13.02	21.83
M2 Pro	112 GFLOPS (8 cores)	6.8 TFLOPS (19 cores)	204.8 GBps	26.04	43.66
M2 Max	112 GFLOPS (8 cores)	13.6 TFLOPS (38 cores)	409.6 GBps	52.08	87.31
M2 Ultra	224 GFLOPS (16 cores)	27.2 TFLOPS (76 cores)	819.2 GBps	104.16	174.62
M3	64.96 GFLOPS (4 cores)	4.1 TFLOPS (10 cores)	102.4 GBps	13.02	21.83
M3 Pro	97.44 GFLOPS (6 cores)	7.4 TFLOPS (18 cores)	153.6 GBps	19.53	32.74
M3 Max	194.88 GFLOPS (12 cores)	16.4 TFLOPS (40 cores)	409.6 GBps	52.08	87.31
M4	70.56 GFLOPS (4 cores)	4.1 TFLOPS	120 GBps	15.26	25.58

References:

https://www.cpu-monkey.com/en/cpu_family-apple_m_series

https://nanoreview.net/

https://en.wikipedia.org/wiki/Apple_M1#Variants

https://en.wikipedia.org/wiki/Apple_M2#Variants

https://en.wikipedia.org/wiki/Apple_M3#Variants

https://en.wikipedia.org/wiki/Apple_M4#Variants

You can further verify the above results in Performance of llama.cpp on Apple Silicon M-series .

Run LLaMA3.1-405B-Instruct with Apple Mac Studio devices combined with Thunderbolt

Example by leafspark/Meta-Llama-3.1-405B-Instruct-GGUF and estimate the maximum tokens per second for three Apple Mac Studio devices combined with Thunderbolt.

Device	CPU FLOPS (Performance Core)	iGPU FLOPS	(V)RAM Bandwidth	Thunderbolt Bandwidth	Role
Apple Mac Studio (M2 Ultra) 0	224 GFLOPS (16 cores)	27.2 TFLOPS (76 cores)	819.2 GBps	40 Gbps	Main
Apple Mac Studio (M2 Ultra) 1	224 GFLOPS (16 cores)	27.2 TFLOPS (76 cores)	819.2 GBps	40 Gbps	RPC Server
Apple Mac Studio (M2 Ultra) 2	224 GFLOPS (16 cores)	27.2 TFLOPS (76 cores)	819.2 GBps	40 Gbps	RPC Server

Get the maximum tokens per second with the following command:

$ # Estimate full offloaded Q4_K_M model.
$ gguf-parser --hf-repo leafspark/Meta-Llama-3.1-405B-Instruct-GGUF --hf-file Llama-3.1-405B-Instruct.Q4_0.gguf/Llama-3.1-405B-Instruct.Q4_0-00001-of-00012.gguf --skip-metadata --skip-architecture --skip-tokenizer --in-short \
  --no-mmap \
  -c 512 \
  --device-metric "224GFLOPS;819.2GBps,27.2TFLOPS;819.2GBps" \
  --rpc host1:port,host2:port \
  --device-metric "27.2TFLOPS;819.2GBps;40Gbps" \
  --device-metric "27.2TFLOPS;819.2GBps;40Gbps" \
  --tensor-split "<Proportions>"

Tensor Split	Apple Mac Studio 0 RAM	Apple Mac Studio 1 VRAM (RPC 0)	Apple Mac Studio 2 VRAM (RPC 1)	Apple Mac Studio 0 VRAM	Q4_0 Max TPS
1,1,1	1.99 GiB	72.74 GiB	71.04 GiB	70.96 GiB	10.26
2,1,1	1.99 GiB	108.26 GiB	54.13 GiB	52.35 GiB	12.27
3,1,1	1.99 GiB	130.25 GiB	42.29 GiB	42.20 GiB	9.41
4,1,1	1.99 GiB	143.78 GiB	35.52 GiB	35.44 GiB	7.86

Full Layers Offload (default)

$ gguf-parser --hf-repo="etemiz/Llama-3.1-405B-Inst-GGUF" --hf-file="llama-3.1-405b-IQ1_M-00019-of-00019.gguf" --skip-metadata --skip-architecture --skip-tokenizer --in-short
+--------------------------------------------------------------------------------------+
| ESTIMATE                                                                             |
+----------------------------------------------+---------------------------------------+
|                      RAM                     |                 VRAM 0                |
+--------------------+------------+------------+----------------+---------+------------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA   |   NONUMA   |
+--------------------+------------+------------+----------------+---------+------------+
|      1 + 0 + 0     | 652.53 MiB | 802.53 MiB |     126 + 1    | 126 GiB | 247.59 GiB |
+--------------------+------------+------------+----------------+---------+------------+

Zero Layers Offload

$ gguf-parser --hf-repo="etemiz/Llama-3.1-405B-Inst-GGUF" --hf-file="llama-3.1-405b-IQ1_M-00019-of-00019.gguf" --skip-metadata --skip-architecture --skip-tokenizer --gpu-layers=0 --in-short
+------------------------------------------------------------------------------------+
| ESTIMATE                                                                           |
+----------------------------------------------+-------------------------------------+
|                      RAM                     |                VRAM 0               |
+--------------------+------------+------------+----------------+--------+-----------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA  |   NONUMA  |
+--------------------+------------+------------+----------------+--------+-----------+
|     1 + 126 + 1    | 126.37 GiB | 126.52 GiB |      0 + 0     |   0 B  | 33.34 GiB |
+--------------------+------------+------------+----------------+--------+-----------+

Specific Layers Offload

$ gguf-parser --hf-repo="etemiz/Llama-3.1-405B-Inst-GGUF" --hf-file="llama-3.1-405b-IQ1_M-00019-of-00019.gguf" --skip-metadata --skip-architecture --skip-tokenizer --gpu-layers=10 --in-short
+------------------------------------------------------------------------------------+
| ESTIMATE                                                                           |
+----------------------------------------------+-------------------------------------+
|                      RAM                     |                VRAM 0               |
+--------------------+------------+------------+----------------+--------+-----------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA  |   NONUMA  |
+--------------------+------------+------------+----------------+--------+-----------+
|     1 + 116 + 1    | 116.64 GiB | 116.78 GiB |     10 + 0     | 10 GiB | 50.39 GiB |
+--------------------+------------+------------+----------------+--------+-----------+

Specific Context Size

By default, the context size retrieved from the model's metadata.

Use --ctx-size to specify the context size.

$ gguf-parser --hf-repo="etemiz/Llama-3.1-405B-Inst-GGUF" --hf-file="llama-3.1-405b-IQ1_M-00019-of-00019.gguf" --skip-metadata --skip-architecture --skip-tokenizer --ctx-size=4096 --in-short
+--------------------------------------------------------------------------------------+
| ESTIMATE                                                                             |
+----------------------------------------------+---------------------------------------+
|                      RAM                     |                 VRAM 0                |
+--------------------+------------+------------+----------------+----------+-----------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |    UMA   |   NONUMA  |
+--------------------+------------+------------+----------------+----------+-----------+
|      1 + 0 + 0     | 404.53 MiB | 554.53 MiB |     126 + 1    | 3.94 GiB | 93.31 GiB |
+--------------------+------------+------------+----------------+----------+-----------+

Enable Flash Attention

By default, LLaMA.cpp disables the Flash Attention.

Enable Flash Attention will reduce the VRAM usage, but it also increases the GPU/CPU usage.

Use --flash-attention to enable the Flash Attention.

Please note that not all models support Flash Attention, if the model does not support, the "FLASH ATTENTION" shows " Disabled" even if you enable it.

$ gguf-parser --hf-repo="etemiz/Llama-3.1-405B-Inst-GGUF" --hf-file="llama-3.1-405b-IQ1_M-00019-of-00019.gguf" --skip-metadata --skip-architecture --skip-tokenizer --flash-attention --in-short
+--------------------------------------------------------------------------------------+
| ESTIMATE                                                                             |
+----------------------------------------------+---------------------------------------+
|                      RAM                     |                 VRAM 0                |
+--------------------+------------+------------+----------------+---------+------------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA   |   NONUMA   |
+--------------------+------------+------------+----------------+---------+------------+
|      1 + 0 + 0     | 620.53 MiB | 770.53 MiB |     126 + 1    | 126 GiB | 215.70 GiB |
+--------------------+------------+------------+----------------+---------+------------+

Disable MMap

By default, LLaMA.cpp loads the model via Memory-Mapped.

For Apple MacOS, Memory-Mapped is an efficient way to load the model, and results in a lower VRAM usage. For other platforms, Memory-Mapped affects the first-time model loading speed only.

Use --no-mmap to disable loading the model via Memory-Mapped.

Please note that some models require loading the whole weight into memory, if the model does not support MMap, the "MMAP LOAD" shows "Not Supported".

$ gguf-parser --hf-repo="etemiz/Llama-3.1-405B-Inst-GGUF" --hf-file="llama-3.1-405b-IQ1_M-00019-of-00019.gguf" --skip-metadata --skip-architecture --skip-tokenizer --no-mmap --in-short
+-------------------------------------------------------------------------------------+
| ESTIMATE                                                                            |
+------------------------------------------+------------------------------------------+
|                    RAM                   |                  VRAM 0                  |
+--------------------+----------+----------+----------------+------------+------------+
| LAYERS (I + T + O) |    UMA   |  NONUMA  | LAYERS (T + O) |     UMA    |   NONUMA   |
+--------------------+----------+----------+----------------+------------+------------+
|      1 + 0 + 0     | 1.98 GiB | 2.13 GiB |     126 + 1    | 213.97 GiB | 247.59 GiB |
+--------------------+----------+----------+----------------+------------+------------+

With Adapter

Use --lora/--control-vector to estimate the usage when loading a model with adapters.

$ gguf-parser --hf-repo="QuantFactory/Meta-Llama-3-8B-Instruct-GGUF" --hf-file="Meta-Llama-3-8B-Instruct.Q5_K_M.gguf" --skip-metadata --skip-architecture --skip-tokenizer --in-short
+-----------------------------------------------------------------------------------+
| ESTIMATE                                                                          |
+----------------------------------------------+------------------------------------+
|                      RAM                     |               VRAM 0               |
+--------------------+------------+------------+----------------+--------+----------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA  |  NONUMA  |
+--------------------+------------+------------+----------------+--------+----------+
|      1 + 0 + 0     | 163.62 MiB | 313.62 MiB |     32 + 1     |  1 GiB | 6.82 GiB |
+--------------------+------------+------------+----------------+--------+----------+

$ # With a LoRA adapter.
$ gguf-parser --hf-repo="QuantFactory/Meta-Llama-3-8B-Instruct-GGUF" --hf-file="Meta-Llama-3-8B-Instruct.Q5_K_M.gguf" --lora-url="https://huggingface.co/ngxson/test_gguf_lora_adapter/resolve/main/lora-Llama-3-Instruct-abliteration-LoRA-8B-f16.gguf" --skip-metadata --skip-architecture --skip-tokenizer --in-short
+-----------------------------------------------------------------------------------+
| ESTIMATE                                                                          |
+----------------------------------------------+------------------------------------+
|                      RAM                     |               VRAM 0               |
+--------------------+------------+------------+----------------+--------+----------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA  |  NONUMA  |
+--------------------+------------+------------+----------------+--------+----------+
|      1 + 0 + 0     | 176.30 MiB | 326.30 MiB |     32 + 1     |  1 GiB | 6.98 GiB |
+--------------------+------------+------------+----------------+--------+----------+

Get Proper Offload Layers

Use --gpu-layers-step to get the proper offload layers number when the model is too large to fit into the GPUs memory.

$ gguf-parser --hf-repo="etemiz/Llama-3.1-405B-Inst-GGUF" --hf-file="llama-3.1-405b-IQ1_M-00019-of-00019.gguf" --skip-metadata --skip-architecture --skip-tokenizer --gpu-layers-step=6 --in-short
+--------------------------------------------------------------------------------------+
| ESTIMATE                                                                             |
+----------------------------------------------+---------------------------------------+
|                      RAM                     |                 VRAM 0                |
+--------------------+------------+------------+----------------+---------+------------+
| LAYERS (I + T + O) |     UMA    |   NONUMA   | LAYERS (T + O) |   UMA   |   NONUMA   |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 126 + 1    | 126.37 GiB | 126.52 GiB |      0 + 0     |   0 B   |  33.34 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 120 + 1    | 120.64 GiB | 120.78 GiB |      6 + 0     |  6 GiB  |  43.68 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 114 + 1    | 114.64 GiB | 114.78 GiB |     12 + 0     |  12 GiB |  53.74 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 108 + 1    | 108.64 GiB | 108.78 GiB |     18 + 0     |  18 GiB |  63.80 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 102 + 1    | 102.64 GiB | 102.78 GiB |     24 + 0     |  24 GiB |  73.86 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 96 + 1     |  96.64 GiB |  96.78 GiB |     30 + 0     |  30 GiB |  83.93 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 90 + 1     |  90.64 GiB |  90.78 GiB |     36 + 0     |  36 GiB |  93.99 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 84 + 1     |  84.64 GiB |  84.78 GiB |     42 + 0     |  42 GiB | 104.05 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 78 + 1     |  78.64 GiB |  78.78 GiB |     48 + 0     |  48 GiB | 114.11 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 72 + 1     |  72.64 GiB |  72.78 GiB |     54 + 0     |  54 GiB | 124.17 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 66 + 1     |  66.64 GiB |  66.78 GiB |     60 + 0     |  60 GiB | 134.23 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 60 + 1     |  60.64 GiB |  60.78 GiB |     66 + 0     |  66 GiB | 144.29 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 54 + 1     |  54.64 GiB |  54.78 GiB |     72 + 0     |  72 GiB | 154.35 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 48 + 1     |  48.64 GiB |  48.78 GiB |     78 + 0     |  78 GiB | 164.42 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 42 + 1     |  42.64 GiB |  42.78 GiB |     84 + 0     |  84 GiB | 174.48 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 36 + 1     |  36.64 GiB |  36.78 GiB |     90 + 0     |  90 GiB | 184.54 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 30 + 1     |  30.64 GiB |  30.78 GiB |     96 + 0     |  96 GiB | 194.60 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 24 + 1     |  24.64 GiB |  24.78 GiB |     102 + 0    | 102 GiB | 204.66 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 18 + 1     |  18.64 GiB |  18.78 GiB |     108 + 0    | 108 GiB | 214.72 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|     1 + 12 + 1     |  12.64 GiB |  12.78 GiB |     114 + 0    | 114 GiB | 225.05 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|      1 + 6 + 1     |  6.64 GiB  |  6.78 GiB  |     120 + 0    | 120 GiB | 235.64 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|      1 + 0 + 1     | 653.08 MiB | 803.08 MiB |     126 + 0    | 126 GiB | 246.24 GiB |
+--------------------+------------+------------+----------------+---------+------------+
|      1 + 0 + 0     | 652.53 MiB | 802.53 MiB |     126 + 1    | 126 GiB | 247.59 GiB |
+--------------------+------------+------------+----------------+---------+------------+

License

MIT

Name		Name	Last commit message	Last commit date
Latest commit History 109 Commits
.github/workflows		.github/workflows
cmd/gguf-parser		cmd/gguf-parser
util		util
.gitattributes		.gitattributes
.gitignore		.gitignore
.golangci.yaml		.golangci.yaml
Dockerfile		Dockerfile
LICENSE		LICENSE
Makefile		Makefile
README.md		README.md
cache.go		cache.go
file.go		file.go
file_architecture.go		file_architecture.go
file_architecture_test.go		file_architecture_test.go
file_estimate.go		file_estimate.go
file_estimate_option.go		file_estimate_option.go
file_estimate_test.go		file_estimate_test.go
file_from_distro.go		file_from_distro.go
file_from_remote.go		file_from_remote.go
file_metadata.go		file_metadata.go
file_metadata_test.go		file_metadata_test.go
file_option.go		file_option.go
file_test.go		file_test.go
file_tokenizer.go		file_tokenizer.go
file_tokenizer_test.go		file_tokenizer_test.go
filename.go		filename.go
filename_test.go		filename_test.go
gen.go		gen.go
gen.stringer.go		gen.stringer.go
ggml.go		ggml.go
go.mod		go.mod
go.sum		go.sum
ollama_model.go		ollama_model.go
ollama_model_option.go		ollama_model_option.go
ollama_model_test.go		ollama_model_test.go
ollama_registry_authenticate.go		ollama_registry_authenticate.go
scalar.go		scalar.go
scalar_test.go		scalar_test.go
zz_generated.ggmltype.stringer.go		zz_generated.ggmltype.stringer.go
zz_generated.gguffiletype.stringer.go		zz_generated.gguffiletype.stringer.go
zz_generated.ggufmagic.stringer.go		zz_generated.ggufmagic.stringer.go
zz_generated.ggufmetadatavaluetype.stringer.go		zz_generated.ggufmetadatavaluetype.stringer.go
zz_generated.ggufversion.stringer.go		zz_generated.ggufversion.stringer.go

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

GGUF Parser

Key Features

Agenda

Notes

Installation

Overview

Parse

Parse Local File

Parse Remote File

Parse From HuggingFace

Parse From ModelScope

Parse From Ollama Library

Parse None Model

Estimate

Across Multiple GPU Devices

Single Host Multiple GPU Devices

Multiple Hosts Multiple GPU Devices

Maximum Tokens Per Second

CPU FLOPS Calculation

Run LLaMA2-7B-Chat with Apple Silicon M-series

Run LLaMA3.1-405B-Instruct with Apple Mac Studio devices combined with Thunderbolt

Full Layers Offload (default)

Zero Layers Offload

Specific Layers Offload

Specific Context Size

Enable Flash Attention

Disable MMap

With Adapter

Get Proper Offload Layers

License

About

Releases 59

Languages

License

gpustack/gguf-parser-go

Folders and files

Latest commit

History

Repository files navigation

GGUF Parser

Key Features

Agenda

Notes

Installation

Overview

Parse

Parse Local File

Parse Remote File

Parse From HuggingFace

Parse From ModelScope

Parse From Ollama Library

Parse None Model

Estimate

Across Multiple GPU Devices

Single Host Multiple GPU Devices

Multiple Hosts Multiple GPU Devices

Maximum Tokens Per Second

CPU FLOPS Calculation

Run LLaMA2-7B-Chat with Apple Silicon M-series

Run LLaMA3.1-405B-Instruct with Apple Mac Studio devices combined with Thunderbolt

Full Layers Offload (default)

Zero Layers Offload

Specific Layers Offload

Specific Context Size

Enable Flash Attention

Disable MMap

With Adapter

Get Proper Offload Layers

License

About

Topics

Resources

License

Stars

Watchers

Forks

Releases 59

Languages