diff options
| author | Nicolas Patry <patry.nicolas@protonmail.com> | 2024-02-13 16:28:56 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-02-13 16:28:56 +0100 |
| commit | c1b418586c9477a85150ce6c15dcfe4c93d3a27d (patch) | |
| tree | daabda27c6689f7ea9db68ceb8f6bd74ebf8f4b1 /candle-core/src/quantized/ggml_file.rs | |
| parent | ad73e93da2cf7311cb5c5bc39250aa335c5f9b76 (diff) | |
| download | candle-c1b418586c9477a85150ce6c15dcfe4c93d3a27d.tar.gz candle-c1b418586c9477a85150ce6c15dcfe4c93d3a27d.tar.bz2 candle-c1b418586c9477a85150ce6c15dcfe4c93d3a27d.zip | |
Fixing quantized llama demo on metal. (#1703)
Diffstat (limited to 'candle-core/src/quantized/ggml_file.rs')
| -rw-r--r-- | candle-core/src/quantized/ggml_file.rs | 3 |
1 files changed, 3 insertions, 0 deletions
diff --git a/candle-core/src/quantized/ggml_file.rs b/candle-core/src/quantized/ggml_file.rs index 38238580..e6f5791c 100644 --- a/candle-core/src/quantized/ggml_file.rs +++ b/candle-core/src/quantized/ggml_file.rs @@ -233,6 +233,7 @@ pub struct Content { pub hparams: HParams, pub vocab: Vocab, pub tensors: HashMap<String, super::QTensor>, + pub device: Device, } impl Content { @@ -252,11 +253,13 @@ impl Content { let (name, tensor) = read_one_tensor(reader, magic, device)?; tensors.insert(name, tensor); } + let device = device.clone(); Ok(Self { magic, hparams, vocab, tensors, + device, }) } |