LLaMA-v2-chinese-alpaca-13B-GGML (ymcui)
Here are the GGML converted and/or quantized models for ymcui's Chinese LLaMA-v2 Alpaca 13B.
!NOTE! The GGML filetype is outdated. Prefer GGUF format going forward.
Explanation of quantisation methods
Click to see details
Methods:
- type-0 (Q4_0, Q5_0, Q8_0) - weights w are obtained from quants q using w = d * q, where d is the block scale.
- type-1 (Q4_1, Q5_1) - weights are given by w = d * q + m, where m is the block minimum
The new methods available are:
- GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
- GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
- GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
- GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
- GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
- GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
This is exposed via llama.cpp quantization types that define various "quantization mixes" as follows:
- LLAMA_FTYPE_MOSTLY_Q2_K - uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors.
- LLAMA_FTYPE_MOSTLY_Q3_K_S - uses GGML_TYPE_Q3_K for all tensors
- LLAMA_FTYPE_MOSTLY_Q3_K_M - uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K
- LLAMA_FTYPE_MOSTLY_Q3_K_L - uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K
- LLAMA_FTYPE_MOSTLY_Q4_K_S - uses GGML_TYPE_Q4_K for all tensors
- LLAMA_FTYPE_MOSTLY_Q4_K_M - uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K
- LLAMA_FTYPE_MOSTLY_Q5_K_S - uses GGML_TYPE_Q5_K for all tensors
- LLAMA_FTYPE_MOSTLY_Q5_K_M - uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K
- LLAMA_FTYPE_MOSTLY_Q6_K- uses 6-bit quantization (GGML_TYPE_Q8_K) for all tensors
Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
|---|---|---|---|---|---|
| llama-v2-chinese-alpaca-13B-Q2_K.ggml | Q2_K | 2 | 5.65 GB | 8.15 GB | smallest, significant quality-loss - not recommended for most purposes |
| llama-v2-chinese-alpaca-13B-Q3_K_S.ggml | Q3_K_S | 3 | 5.81 GB | 8.31 GB | very small, high quality-loss |
| llama-v2-chinese-alpaca-13B-Q3_K_M.ggml | Q3_K_M | 3 | 6.46 GB | 7.96 GB | very small, high quality-loss |
| llama-v2-chinese-alpaca-13B-Q3_K_L.ggml | Q3_K_L | 3 | 7.08 GB | 9.58 GB | small, substantial quality-loss |
| llama-v2-chinese-alpaca-13B-Q4_0.ggml | Q4_0 | 4 | 7.53 GB | 10.03 GB | legacy; small, very high quality-loss - prefer using Q3_K_M |
| llama-v2-chinese-alpaca-13B-Q4_1.ggml | Q4_1 | 4 | 8.34 GB | 10.84 GB | legacy; small, very high quality-loss - prefer using Q3_K_M |
| llama-v2-chinese-alpaca-13B-Q4_K_S.ggml | Q4_K_S | 4 | 7.53 GB | 10.03 GB | small, greater quality-loss |
| llama-v2-chinese-alpaca-13B-Q4_K_M.ggml | Q4_K_M | 4 | 8.03 GB | 10.53 GB | medium, balanced quality - recommended |
| llama-v2-chinese-alpaca-13B-Q5_0.ggml | Q5_0 | 5 | 9.15 GB | 11.65 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| llama-v2-chinese-alpaca-13B-Q5_1.ggml | Q5_1 | 5 | 9.96 GB | 12.46 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| llama-v2-chinese-alpaca-13B-Q5_K_S.ggml | Q5_K_S | 5 | 9.15 GB | 11.65 GB | large, low quality-loss - recommended |
| llama-v2-chinese-alpaca-13B-Q5_K_M.ggml | Q5_K_M | 5 | 9.41 GB | 11.91 GB | large, very low quality-loss - recommended |
| llama-v2-chinese-alpaca-13B-Q6_K.ggml | Q6_K | 6 | 10.9 GB | 13.4 GB | very large, extremely low quality-loss |
| llama-v2-chinese-alpaca-13B-Q8_0.ggml | Q8_0 | 8 | 14 GB | 16.5 GB | very large, extremely low quality-loss - not recommended |
| llama-v2-chinese-alpaca-13B-f16.ggml | f16 | 16 | 26.5 GB | 29 GB | very large, almost no quality-loss - not recommended |
Model Sources
- Repository: [https://github.com/ymcui/Chinese-LLaMA-Alpaca-2]