Files
llama.cpp/tools/mtmd/models/inkling2v.cpp
T
Piotr Wilkin (ilintar)andXuan Son Nguyen 5c008e0008 feat: add Inkling 2 support (text + vision) (#32217)
* convert : add Inkling2 text GGUF conversion

Assisted-by: Codex

* llama : implement Inkling2 text inference

Assisted-by: Codex

* tests : cover Inkling2 text support

Assisted-by: Codex

* conversion : dequantize packed Inkling2 experts in row chunks

Decode per-layer NVFP4 and MXFP8 schemes, validate companions, and use stored scale shapes with a warning for the declared MXFP8 [32,32] grid. Reorder packed w13 rows and their scale rows together. Keep normal GGUF outtype selection and 3D expert tensors.

Stream 128 rows per expert through LazyChunkedTensor. Real w13 (384x4608x4608) needs a 2.25 MiB F32 chunk instead of a 30.375 GiB expanded layer. Real-shaped sparse probes peak below 5.8 MiB of Python/NumPy chunk allocations, plus 5.2 MiB of callback metadata: budget about 16 MiB per projection including mapped pages, excluding runtime, other tensors and OS cache. Full-model chunk metadata is roughly 370 MiB.

Packed synthetic GGUF weights match the independent pinned vLLM decode bit-for-bit. CPU prefill/decode logits remain within 3.1e-6 absolute error, with 100% argmax agreement. Float fixtures, original-name conversion and CPU architecture checks pass. Real checkpoint conversion remains untested.

Assisted-by: Codex

* convert : add Inkling2 vision mmproj support

Assisted-by: Codex

* mtmd : implement Inkling2 still-image encoder

Assisted-by: Codex

* inkling2 : scope image loading

Assisted-by: Codex

* inkling2 : enable flash attention with folded bias and sinks

Assisted-by: Codex

* cuda : route per-head flash attention masks through vec

Assisted-by: Codex

* mtmd : support Inkling2 temporal frame means

Assisted-by: Codex

* style : restore shared image comment

Assisted-by: Codex

* tests : enlarge Inkling2 context fixture

Assisted-by: Codex

* inkling2 : load per-layer expert feed-forward widths

Assisted-by: Codex

* tests : add Inkling2 Metal fusion baselines

Assisted-by: Codex

* inkling2 : group and align registry entries

Assisted-by: Fable

* convert : regenerate the Inkling2 tokenizer hash with the update script

Assisted-by: Fable

* conversion : revert the print_registered_models change

Assisted-by: Fable

* conversion : sort the Inkling2 model map entries

Assisted-by: Fable

* convert : map Inkling2 vision checkpoint names in the tensor table

Assisted-by: Fable

* convert : map Inkling2 text checkpoint names in the tensor table

Assisted-by: Fable

* tests : document the Inkling2 packed expert conversion test

Assisted-by: Fable

* tests : cover Inkling2 checkpoint and transformers tensor names

Assisted-by: Fable

* llama : place Inkling2 fields and list entries by file convention

Assisted-by: Fable

* gguf-py : align Inkling2 tensor mapping comments

Assisted-by: Fable

* mtmd : place Inkling2 entries by file convention

Assisted-by: Fable

* tests : place Inkling2 cases after the existing ones

Assisted-by: Fable

* cuda : index per-head flash attention masks in the tile and mma kernels

Assisted-by: Claude
Claude-Session: https://claude.ai/code/session_01GbYSza1yxgmTz8wdr6uV8Q

* inkling2 : project the relative bias before the gather, drop arch branches

Assisted-by: Claude
Claude-Session: https://claude.ai/code/session_01GbYSza1yxgmTz8wdr6uV8Q

* convert : require Inkling2 hparams, add inkling2 pre-tokenizer

Assisted-by: Claude
Claude-Session: https://claude.ai/code/session_01GbYSza1yxgmTz8wdr6uV8Q

* mtmd : cap Inkling2 image size, drop video frame averaging

Assisted-by: Claude
Claude-Session: https://claude.ai/code/session_01GbYSza1yxgmTz8wdr6uV8Q

* inkling2 : fix CI build, Meta device assert and Windows conversion test

Assisted-by: Claude

* tests : use the default sliding window for the Inkling2 test model

Assisted-by: Claude

* simplify mtmd part

* give _v arch for vision

* remove alpha channel handling, it should be a follow up

* explicit rule about audio + vision models

* add audio support

* rm audio

* rm gguf mtmd rope base

* address review comment

* nits

* do not normalize embd from mtmd

---------

Co-authored-by: Xuan Son Nguyen <son@huggingface.co>
2026-11-15 00:51:31 +01:00

34 lines
1.4 KiB
C++

#include "models.h"
// ViT with a CLS token, learned position embeddings and 2D RoPE; the patches are merged before the projector
ggml_cgraph * mtmd_graph_inkling2v::build() {
const int n_pos = n_patches + 1;
auto * pos_w = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, n_pos);
ggml_set_name(pos_w, "pos_w");
ggml_set_input(pos_w);
auto * pos_h = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, n_pos);
ggml_set_name(pos_h, "pos_h");
ggml_set_input(pos_h);
auto * cur = build_inp();
cb(cur, "patch_embd", -1);
cur = ggml_add(ctx0, cur, resize_position_embeddings(GGML_SCALE_MODE_BILINEAR));
auto * cls = ggml_add(ctx0, model.class_embedding, model.class_position_embedding);
cur = ggml_concat(ctx0, cls, cur, 1);
cb(cur, "embeddings", -1);
auto add_pos = [&](ggml_tensor * x, const mtmd_layer &) {
return build_rope_2d(ctx0, x, pos_w, pos_h, hparams.rope_theta, false);
};
cur = build_vit(cur, n_pos, NORM_TYPE_NORMAL, hparams.ffn_op, nullptr, add_pos);
cb(cur, "encoder_out", -1);
// remove CLS token
cur = ggml_view_2d(ctx0, cur, n_embd, n_patches, cur->nb[1], cur->nb[1]);
cur = build_patch_merge_permute(cur, hparams.n_merge);
cur = build_ffn(cur, model.mm_1_w, model.mm_1_b, nullptr, nullptr, model.mm_2_w, model.mm_2_b, FFN_GELU_ERF, -1);
cb(cur, "projected", -1);
ggml_build_forward_expand(gf, cur);
return gf;
}