candle
candle copied to clipboard
Qwen/Qwen2-7B doesn't work properly in the example qwen
the Qwen/Qwen2-1.5B can work correct in the example, but Qwen/Qwen2-7B can't.
(base) lyn@A100DEV:~/workspace/candle/candle-examples$ cargo run --release --features cuda --example qwen -- --model 2-7b --prompt "Hello\n" Finished release [optimized] target(s) in 0.17s Running /disk/lyn/workspace/candle/target/release/examples/qwen --model 2-7b --prompt 'Hello\n' avx: true, neon: false, simd128: false, f16c: true temp: 0.00 repeat-penalty: 1.10 repeat-last-n: 64 retrieved the files in 17.702388ms loaded the model in 2.506053906s Hello\n规范、、、、、、、 Orient Vall gu gap FourierbrtooltooltoolCharlesuketooltoolaurustooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltooltoolaurustoolrm哪些
The model output is obviously incorrect.
the version: commit 242e006bbb26ff12581b3c04bfd069996fe1f6bb (HEAD -> main, origin/main, origin/HEAD) Author: Jeroen Vlek [email protected] Date: Mon Jun 24 19:12:52 2024 +0200