mirror of
https://github.com/ollama/ollama.git
synced 2026-10-14 19:34:53 +00:00
d9dec1509d
* Reduce model lookup and MLX decision request overhead Avoid decoding unrelated manifests when resolving existing model names. Reuse small Metal scratch buffers between decision requests, releasing larger working sets and clearing scratch after scoring errors. * address comments |
||
|---|---|---|
| .. | ||
| batch | ||
| cache | ||
| model | ||
| nn | ||
| sample | ||
| tokenizer | ||
| xgrammar | ||
| cache_trie.go | ||
| cache_trie_test.go | ||
| client.go | ||
| client_score.go | ||
| client_score_test.go | ||
| client_test.go | ||
| dflash.go | ||
| dflash_test.go | ||
| grammar.go | ||
| grammar_mask_test.go | ||
| grammar_test.go | ||
| media.go | ||
| media_test.go | ||
| mtp.go | ||
| mtp_test.go | ||
| pipeline.go | ||
| pipeline_test.go | ||
| prefix_cache.go | ||
| prefix_cache_scenario_test.go | ||
| prefix_cache_test.go | ||
| runner.go | ||
| score.go | ||
| score_hidden_test.go | ||
| score_test.go | ||
| server.go | ||
| speculate.go | ||
| speculate_depth.go | ||
| speculate_depth_test.go | ||
| speculate_stats.go | ||
| status_memory.go | ||
| status_memory_test.go | ||
| utf8_buffer.go | ||
| utf8_buffer_test.go | ||
| verify.go | ||
| verify_test.go | ||