---

language:

• en
• de
• fr
• it
• pt
• hi
• es
• th

tags:

• quantized
• 2-bit
• 3-bit
• GGUF
• text-generation
• text-generation

model_name: Meta-Llama-3.1-405B-Instruct-GGUF base_model: meta-llama/Meta-Llama-3.1-405B-Instruct inference: false model_creator: meta-llama pipeline_tag: text-generation quantized_by: MaziyarPanahi license: llama3.1

---

MaziyarPanahi/Meta-Llama-3.1-405B-Instruct-GGUF

• Model creator: meta-llama
• Original model: meta-llama/Meta-Llama-3.1-405B-Instruct

Description

MaziyarPanahi/Meta-Llama-3.1-405B-Instruct-GGUF contains GGUF format model files for meta-llama/Meta-Llama-3.1-405B-Instruct.

Sample

llama.cpp/llama-cli -m Meta-Llama-3.1-405B-Instruct.Q2_K.gguf-00001-of-00009.gguf -p "write 10 sentences ending with the word apple." -n 1024 -t 40

systeminfo: nthreads = 40 / 80 | AVX = 1 | AVXVNNI = 0 | AVX2 = 1 | AVX512 = 0 | AVX512VBMI = 0 | AVX512VNNI = 0 | AVX512BF16 = 0 | FMA = 1 | NEON = 0 | SVE = 0 | ARMFMA = 0 | F16C = 1 | FP16VA = 0 | WASMSIMD = 0 | BLAS = 0 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMULINT8 = 0 | LLAMAFILE = 1 |
sampling:
        repeatlastn = 64, repeatpenalty = 1.000, frequencypenalty = 0.000, presence_penalty = 0.000
        topk = 40, tfsz = 1.000, topp = 0.950, minp = 0.050, typical_p = 1.000, temp = 0.800
        mirostat = 0, mirostatlr = 0.100, mirostatent = 5.000
sampling order:
CFG -> Penalties -> topk -> tfsz -> typicalp -> topp -> min_p -> temperature
generate: nctx = 131072, nbatch = 2048, npredict = 1024, nkeep = 1

write 10 sentences ending with the word apple.
I love to eat a crunchy, juicy apple.
The teacher gave the student a shiny, red apple.
The farmer plucked a ripe, delicious apple.
My favorite snack is a sweet, tasty apple.
The child picked a fresh, green apple.
The cafeteria served a healthy, sliced apple.
The vendor sold a crisp, autumn apple.
The artist painted a still life with a golden apple.
The baby took a big bite of a soft, mealy apple.
The family enjoyed a basket of fresh, orchard apple. [end of text]


llamaprinttimings:        load time = 1068588.13 ms
llamaprinttimings:      sample time =    2262.60 ms /   136 runs   (   16.64 ms per token,    60.11 tokens per second)
llamaprinttimings: prompt eval time =  339484.02 ms /    11 tokens (30862.18 ms per token,     0.03 tokens per second)
llamaprinttimings:        eval time = 33458013.45 ms /   135 runs   (247837.14 ms per token,     0.00 tokens per second)
llamaprinttimings:       total time = 33800561.08 ms /   146 tokens
Log end

About GGUF

GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.

Here is an incomplete list of clients and libraries that are known to support GGUF:

• llama.cpp. The source project for GGUF. Offers a CLI and a server option.
• llama-cpp-python, a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
• LM Studio, an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration. Linux available, in beta as of 27/11/2023.
• text-generation-webui, the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
• KoboldCpp, a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
• GPT4All, a free and open source local running GUI, supporting Windows, Linux and macOS with full GPU accel.
• LoLLMS Web UI, a great web UI with many interesting and unique features, including a full model library for easy model selection.
• Faraday.dev, an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
• candle, a Rust ML framework with a focus on performance, including GPU support, and ease of use.
• ctransformers, a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server. Note, as of time of writing (November 27th 2023), ctransformers has not been updated in a long time and does not support many recent models.

Special thanks

🙏 Special thanks to Georgi Gerganov and the whole team working on llama.cpp for making all of this possible.