diff options
| author | Pedro Cuenca <pedro@huggingface.co> | 2024-04-21 13:50:41 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-04-21 14:50:41 +0300 |
| commit | b97bc3966e852adb626c90be64fd48282800f504 (patch) | |
| tree | 178656d15821205889fa03ec603c7327facbb265 /examples/beam-search | |
| parent | b8109bc0139f15a5b321909f47510b89dca47ffc (diff) | |
llama : support Llama 3 HF conversion (#6745)
* Support Llama 3 conversion
The tokenizer is BPE.
* style
* Accept suggestion
Co-authored-by: Sourab Mangrulkar <13534540+pacman100@users.noreply.github.com>
* llama : add llama_token_is_eog()
ggml-ci
* llama : auto-detect more EOT tokens when missing in KV data
* convert : replacing EOS token is a hack
* llama : fix codegemma EOT token + add TODOs
* llama : fix model type string for 8B model
---------
Co-authored-by: Sourab Mangrulkar <13534540+pacman100@users.noreply.github.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Diffstat (limited to 'examples/beam-search')
| -rw-r--r-- | examples/beam-search/beam-search.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/examples/beam-search/beam-search.cpp b/examples/beam-search/beam-search.cpp index 866c6d7a..3d34378a 100644 --- a/examples/beam-search/beam-search.cpp +++ b/examples/beam-search/beam-search.cpp @@ -47,7 +47,7 @@ struct beam_search_callback_data { // In this case, end-of-beam (eob) is equivalent to end-of-sentence (eos) but this need not always be the same. // For example, eob can be flagged due to maximum token length, stop words, etc. static bool is_at_eob(const beam_search_callback_data & callback_data, const llama_token * tokens, size_t n_tokens) { - return n_tokens && tokens[n_tokens-1] == llama_token_eos(llama_get_model(callback_data.ctx)); + return n_tokens && llama_token_is_eog(llama_get_model(callback_data.ctx), tokens[n_tokens-1]); } // Function matching type llama_beam_search_callback_fn_t. |