diff --git a/serverAppConfigs-docs/LLM/Radiance Operations.md b/serverAppConfigs-docs/LLM/Radiance Operations.md index 0afff85..8037d26 100644 --- a/serverAppConfigs-docs/LLM/Radiance Operations.md +++ b/serverAppConfigs-docs/LLM/Radiance Operations.md @@ -181,7 +181,8 @@ refusal**, not an engine fault: read the journal line and the argv, not the GPU. proposals held, KV bytes and the tokens they address, stream misses and bytes, prefix hits and evictions, preemptions and queue depth. `/stats`, `/graph`, `/sessions` back the dashboard. `scripts/statsdump.py` records `/stats` to JSONL with - time and gap, for a machine with no scraper. + time and gap, for a machine with no scraper (its key comes from `--api-key` or + `RADIANCE_API_KEY`). - `RADIANCE_PROFILE_EVERY`, `RADIANCE_DUMP_BUF`/`RADIANCE_DUMP_DIR`, `RADIANCE_OP_ORACLE`(+) and the `RADIANCE_DEBUG_*` family are debug instruments; they are not deployment settings. diff --git a/serverAppConfigs-docs/LLM/Radiance Tools.md b/serverAppConfigs-docs/LLM/Radiance Tools.md index c2d1204..9074d62 100644 --- a/serverAppConfigs-docs/LLM/Radiance Tools.md +++ b/serverAppConfigs-docs/LLM/Radiance Tools.md @@ -255,7 +255,7 @@ eight-rank all-reduce shard weights and times `lockstep`/`ring`/`split` forms at ``` rad-tokcheck [options] ... - --hf-dir DIR compare against HuggingFace ids in DIR/.ids + --hf-dir DIR compare against HuggingFace ids in DIR/.ids (little-endian int32, encode(text, add_special_tokens=False)) --no-ref skip the reference encoder (slow) --user encode with parse_special off