DFlash polishing checkpoint weights update

README.md

@@ -20,13 +20,14 @@ base_model:
 
 <br>
 
-# Laguna-S-2.1-DFlash
+# Laguna-S-2.1-DFlash-polishing
 
-DFlash speculator (draft model) for
+DFlash speculator (drafter) for
 [Laguna S 2.1](https://huggingface.co/poolside/Laguna-S-2.1), bf16.
 
 - Architecture: `DFlashLagunaForCausalLM` (6 sliding-attention layers, block_size 16).
 - Shares token embedding + lm_head with the target; `draft_vocab_size == vocab_size` (no d2t/t2d).
+- Source checkpoint: `s3://poolside.titan.checkpoints.us-east-2/adam/checkpoints/ft_sft_e0630_rhiemann_baseline_titan_sft_training/0008400/DFlash_Stage2/25000`
 - Loads under vLLM (native `laguna_dflash`) and TRT-LLM (pytorch DFlash backend) as the draft model in a speculative config.
 
 ## Usage
@@ -50,7 +51,7 @@ python -m sglang.launch_server --model-path poolside/Laguna-S-2.1 \
 
 ### llama.cpp (GGUF)
 
-A llama.cpp conversion of this draft model (`laguna-s-2.1-DFlash-BF16.gguf`) is
+A llama.cpp conversion of this drafter (`laguna-s-2.1-DFlash-BF16.gguf`) is
 published in
 [poolside/Laguna-S-2.1-GGUF](https://huggingface.co/poolside/Laguna-S-2.1-GGUF)
 alongside the target GGUFs. It embeds the target tokenizer and the DFlash metadata
@@ -64,16 +65,16 @@ cd llama.cpp && cmake -B build && cmake --build build -j
 ./build/bin/llama-server \
   -m laguna-s-2.1-Q4_K_M.gguf \
   -md laguna-s-2.1-DFlash-BF16.gguf \
-  --spec-type draft-dflash --spec-draft-n-max 7 \
+  --spec-type draft-dflash --spec-draft-n-max 15 \
   -fa on --jinja --port 8000
 ```
 
 `--spec-draft-n-max` is clamped to the trained block size (15 draft tokens + 1).
 
 > [!NOTE]
-> Requires Poolside's llama.cpp fork, branch
+> Requires poolside's llama.cpp fork, branch
 > [`laguna`](https://github.com/poolsideai/llama.cpp/tree/laguna). Upstream
 > llama.cpp ships the generic DFlash framework but not the Laguna decoder
-> contract this draft model needs, and upstream PR
+> contract this drafter needs, and upstream PR
 > [ggml-org/llama.cpp#25165](https://github.com/ggml-org/llama.cpp/pull/25165)
 > covers the target architecture only.