Views
No views yet
poolside/Laguna-XS.2.
Converted and tested with a patched fork of llama.cpp that implements LLM_ARCH_LAGUNA from scratch.Requires the custom fork. Upstream llama.cpp does not support this architecture. Fork: https://github.com/linuxid10t/llama.cpp-add-lagunaThe same fork also converts and runs the Laguna-M.1 sibling (~226B, all-full-attention, per-element attention gate) — see its own repo.
⚠️ These files were reconverted2026-06-20from the latest publishedpoolside/Laguna-XS.2(config.jsonlast modified 2026-06-18). The previous GGUFs in this repo (a) lacked therope.dimension_countkey, so the global layers silently ran full rotary instead of half (partial_rotary_factor = 0.5), and (b) were produced before several converter fixes. The current files correct this. Verify the key is present (head_dim 128 → expect64):python3 gguf-py/scripts/gguf_dump.py Laguna-XS.2-Q4_K_M.gguf | grep rope.dimension_count
| File | Quant | Size | Notes |
|---|---|---|---|
Laguna-XS.2-f16.gguf | f16 | ~67 GB | Full precision, reference |
Laguna-XS.2-Q4_K_M.gguf | Q4_K_M | ~20 GB | Recommended for most users |
git clone https://github.com/linuxid10t/llama.cpp-add-laguna
cd llama.cpp-add-laguna && cmake -B build && cmake --build build -j$(nproc)
./build/bin/llama-cli \
-m Laguna-XS.2-Q4_K_M.gguf \
--ctx-size 262144 \
--temp 0 \
-p "The capital of France is"# Thinking on (default) — model prefills <think> and generates a reasoning trace
./build/bin/llama-cli -m Laguna-XS.2-Q4_K_M.gguf -cnv --ctx-size 32768
# Thinking off — direct answer, no reasoning trace
./build/bin/llama-cli -m Laguna-XS.2-Q4_K_M.gguf -cnv --ctx-size 32768 --reasoning off</assistant>, token 24), and streaming
parser are all handled automatically — no manual prompt engineering needed.config.json (full-attention layers = global, sliding-attention layers = SWA):| Property | Value |
|---|---|
| Parameters | 33B total, ~3B active per token (A3B) |
| Layers | 40 (10 global + 30 SWA, 3:1 pattern) |
| Attention heads | 48 (global layers) / 64 (SWA layers) |
| KV heads | 8 (GQA) |
| Head dim | 128 |
| Q/K norm | RMSNorm per head |
| Attention gate | Per-head softplus gate on SDPA output, applied on all layers |
| SWA window | 512 tokens |
| Experts | 256 routed (top-8) + 1 shared |
| Dense layers | Layer 0 only |
| MoE router | Sigmoid + e_score_correction_bias (added at selection, not routing) |
| Routing scale | moe_routed_scaling_factor = 2.5, L1-normalized weights |
| Global RoPE | YaRN: base 500K, factor 64, original_max 4096, β_fast 64, β_slow 1, partial_rotary 0.5 |
| SWA RoPE | Default: base 10K, full rotary |
| Context length | 262,144 tokens |
self_attn.g_proj that projects the hidden state to
one scalar per head (num_heads outputs). This is softplus-gated and multiplied,
broadcast across head_dim, into the SDPA output before o_proj. (XS.2 is per-head gating.
The same graph builder does Laguna-M.1's per-element variant, selected by the
attention.gate_per_head GGUF key — detected from the actual g_proj tensor shape by the
converter, since config gating is a bool rather than a mode string.)partial_rotary_factor = 0.5); SWA layers use full rotary. The converter writes this explicitly
as rope.dimension_count (global) and rope.dimension_count_swa (SWA), derived from each rope
config's partial_rotary_factor, and the model reads the per-layer count via n_rot(il) at
graph-build time. (For head_dim 128 this is 64 global / 128 SWA. This replaces an earlier
hardcoded n_rot_full /= 2.)llama_hparams / llama_cparams and selects at graph
build time via hparams.is_swa(il).e_score_correction_bias that is added only during top-k selection, not during weight
computation. Routing weights are L1-normalized before scaling by moe_routed_scaling_factor.<think> as the generation prompt prefix
when thinking is enabled. The streaming parser is patched to treat this as a delimiter-style
boundary (not a generated token), so the <think> tag never appears in streaming output.</assistant> is token 24, a regular vocabulary token (not a special token).
The fork adds it to antiprompt so the stop-word erase logic strips it before output."The capital of France is" → " Paris.\nThe capital of Germany is" ✓SWORDFISH99) — correctly retrieved (exercises global YaRN layers) ✓Q4_K_M; the half-rotary global layers now match the
reference config.)