mirror of
https://github.com/ggerganov/llama.cpp
synced 2026-04-02 05:15:38 +02:00
55c1b2a3bb
* iq1_m: basics * iq1_m: basics-2 * iq1_m: CUDA dequantize works Very 1st shot I get PPL = 9.76 for LLaMA-v2-7B. * iq1_m: separate shifts for each group of 8 in a block We get PPL(LLaMA-v2-7B ) = 9.2810 PPL(LLaMA-v2-13B) = 6.8105 Not bad, but slightly higher than sqrt(PPL(IQ1_S) * PPL(IQ2_XXS)) which is the expected outcome given that IQ1_M is halfway between IQ1_S and IQ2_XXS in terms of bpw. From this, we would expect PPL = 9.14 for LLaMA-v2-7B PPL = 6.63 for LLaMA-v2-13B * iq1_m: go to 3-bit scales There is slight increase in PPL, but the 0.0625 bpw reduction in size is totally worth it. We now have PPL(LLaMA-v2-7B ) = 9.4469 at 1.96 bpw PPL(LLaMA-v2-13B) = 6.8717 at 1.93 bpw PPL(LLaMA-v2-70B) = 4.8568 at 1.85 bpw * iq1_m: scalar dot product * iq1_m: AVX2 dot product * iq1_m: very slightly faster AVX2 dot product * iq1_m: ARM_NEON dot product Works, but very slow (10.5 t/s) * iq1_m: Metal - dequantize works, dot product does not * iq1_m: Metal now works About the same performance as iq1_s. * iq1_m: minor * iq1_m: checking pure iq1_m quantization It is pretty bad: PPL(LLaMA-v2-7B) = 34 if we quantize output.weight with Q4_K. * iiq1_m: slightly faster ARM_NEON dot product 10.5 t/s -> 11.65 t/s * iq1_m: faster ARM_NEON dot product 11.65 t/s -> 14.9 t/s * iq1_m: another minor ARM_NEON dot product improvement 14.9 -> 15.0 t/s * iq1_m: small PPL improvement via super-block scale adjustment After quantizing block scales redo the super-block scale fit. PPL(LLaMA-v2-7B ) = 9.3346 PPL(LLaMA-v2-13B) = 6.8419 PPL(LLaMA-v2-70B) = 4.8294 PPL(Mistral-7B ) = 8.1624 * iq1_m: adapt to CUDA refactoring * iq1_m: remove unused variable We have progressed to warnings being errors. * iq1_m: add to backend-ops tests * iq1_m: fix Windows ARM * iq1_m: use common definition of iq1m_scale_t * cuda: assert -> NO_DEVICE_CODE * iq1_M: PR comments --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com> |
||
|---|---|---|
| .. | ||
| baby-llama | ||
| batched | ||
| batched-bench | ||
| batched.swift | ||
| beam-search | ||
| benchmark | ||
| convert-llama2c-to-ggml | ||
| embedding | ||
| export-lora | ||
| finetune | ||
| gguf | ||
| gguf-split | ||
| gritlm | ||
| imatrix | ||
| infill | ||
| jeopardy | ||
| llama-bench | ||
| llama.android | ||
| llama.swiftui | ||
| llava | ||
| lookahead | ||
| lookup | ||
| main | ||
| main-cmake-pkg | ||
| parallel | ||
| passkey | ||
| perplexity | ||
| quantize | ||
| quantize-stats | ||
| retrieval | ||
| save-load-state | ||
| server | ||
| simple | ||
| speculative | ||
| sycl | ||
| tokenize | ||
| train-text-from-scratch | ||
| alpaca.sh | ||
| base-translate.sh | ||
| chat-13B.bat | ||
| chat-13B.sh | ||
| chat-persistent.sh | ||
| chat-vicuna.sh | ||
| chat.sh | ||
| CMakeLists.txt | ||
| gpt4all.sh | ||
| json-schema-pydantic-example.py | ||
| json-schema-to-grammar.py | ||
| llama2-13b.sh | ||
| llama2.sh | ||
| llama.vim | ||
| llm.vim | ||
| make-ggml.py | ||
| Miku.sh | ||
| pydantic_models_to_grammar.py | ||
| pydantic-models-to-grammar-examples.py | ||
| reason-act.sh | ||
| regex-to-grammar.py | ||
| server-embd.py | ||
| server-llama2-13B.sh | ||
| ts-type-to-grammar.sh | ||