README.md

metadata

license: llama2
datasets:
  - RUCKBReasoning/TableLLM-SFT
language:
  - en
tags:
  - Table
  - QA
  - Code
quantized_by: bartowski
pipeline_tag: text-generation

Llamacpp Quantizations of TableLLM-13b

Using llama.cpp release b2589 for quantization.

Original model: https://huggingface.co/RUCKBReasoning/TableLLM-13b

Download a file (not the whole branch) from below:

Filename	Quant type	File Size	Description
TableLLM-13b-Q8_0.gguf	Q8_0	13.83GB	Extremely high quality, generally unneeded but max available quant.
TableLLM-13b-Q6_K.gguf	Q6_K	10.67GB	Very high quality, near perfect, recommended.
TableLLM-13b-Q5_K_M.gguf	Q5_K_M	9.23GB	High quality, very usable.
TableLLM-13b-Q5_K_S.gguf	Q5_K_S	8.97GB	High quality, very usable.
TableLLM-13b-Q5_0.gguf	Q5_0	8.97GB	High quality, older format, generally not recommended.
TableLLM-13b-Q4_K_M.gguf	Q4_K_M	7.86GB	Good quality, uses about 4.83 bits per weight.
TableLLM-13b-Q4_K_S.gguf	Q4_K_S	7.42GB	Slightly lower quality with small space savings.
TableLLM-13b-IQ4_NL.gguf	IQ4_NL	7.41GB	Decent quality, similar to Q4_K_S, new method of quanting,
TableLLM-13b-IQ4_XS.gguf	IQ4_XS	7.01GB	Decent quality, new method with similar performance to Q4.
TableLLM-13b-Q4_0.gguf	Q4_0	7.36GB	Decent quality, older format, generally not recommended.
TableLLM-13b-Q3_K_L.gguf	Q3_K_L	6.92GB	Lower quality but usable, good for low RAM availability.
TableLLM-13b-Q3_K_M.gguf	Q3_K_M	6.33GB	Even lower quality.
TableLLM-13b-IQ3_M.gguf	IQ3_M	5.98GB	Medium-low quality, new method with decent performance.
TableLLM-13b-IQ3_S.gguf	IQ3_S	5.65GB	Lower quality, new method with decent performance, recommended over Q3 quants.
TableLLM-13b-Q3_K_S.gguf	Q3_K_S	5.65GB	Low quality, not recommended.
TableLLM-13b-Q2_K.gguf	Q2_K	4.85GB	Extremely low quality, not recommended.

Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski