vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-13 18:39:48 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

Support diffusion models: Add Dream 7B (#14644)

Browse source 
* Support diffusion models: Add Dream 7B

* Move diffusion to examples

* Move stuff to examples. Add patch to not use kv-cache

* Address review comments

* Make sampling fast

* llama: remove diffusion functions

* Add basic timings + cleanup

* More cleanup

* Review comments: better formating, use LOG instead std::cerr, re-use batch, use ubatch instead of max_length

* fixup!

* Review: move everything to diffusion-cli for now
This commit is contained in:
Aman Gupta 2025-07-16 20:03:51 +08:00 • committed by GitHub
parent 64978340b0
commit ab14019821
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
13 changed files with 804 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

126
src/llama-model.cpp
View file

@ -849,6 +849,21 @@ void llama_model::load_hparams(llama_model_loader & ml) {
default: type = LLM_TYPE_UNKNOWN;
}
} break;
case LLM_ARCH_DREAM:
{
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
// Dream models are primarily 7B with 28 layers
switch (hparams.n_layer) {
case 28:
type = LLM_TYPE_7B;
break;
default:
type = LLM_TYPE_UNKNOWN;
}
// Set non-causal attention for diffusion models
hparams.causal_attn = false;
}
break;
case LLM_ARCH_QWEN2MOE:
{
ml.get_key(LLM_KV_EXPERT_FEED_FORWARD_LENGTH, hparams.n_ff_exp, false);
@ -2670,6 +2685,7 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
} break;
case LLM_ARCH_QWEN2:
case LLM_ARCH_QWEN2VL:
case LLM_ARCH_DREAM:
{
tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0);
@ -7756,6 +7772,109 @@ struct llm_build_qwen2 : public llm_graph_context {
}
};
struct llm_build_dream : public llm_graph_context {
llm_build_dream(const llama_model & model, const llm_graph_params & params, ggml_cgraph * gf) :
llm_graph_context(params) {
//copied from qwen2
const int64_t n_embd_head = hparams.n_embd_head_v;
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
GGML_ASSERT(n_embd_head == hparams.n_rot);
ggml_tensor * cur;
ggml_tensor * inpL;
inpL = build_inp_embd(model.tok_embd);
// inp_pos - contains the positions
ggml_tensor * inp_pos = build_inp_pos();
auto * inp_attn = build_attn_inp_no_cache();
ggml_tensor * inp_out_ids = build_inp_out_ids();
for (int il = 0; il < n_layer; ++il) {
ggml_tensor * inpSA = inpL;
// norm
cur = build_norm(inpL, model.layers[il].attn_norm, NULL, LLM_NORM_RMS, il);
cb(cur, "attn_norm", il);
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, nullptr, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
Kcur = ggml_rope_ext(ctx0, Kcur, inp_pos, nullptr, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn, gf, model.layers[il].wo, model.layers[il].bo, Qcur, Kcur, Vcur, nullptr,
nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
}
ggml_tensor * ffn_inp = ggml_add(ctx0, cur, inpSA);
cb(ffn_inp, "ffn_inp", il);
// feed-forward network
cur = build_norm(ffn_inp, model.layers[il].ffn_norm, NULL, LLM_NORM_RMS, il);
cb(cur, "ffn_norm", il);
cur = build_ffn(cur, model.layers[il].ffn_up, NULL, NULL, model.layers[il].ffn_gate, NULL, NULL,
model.layers[il].ffn_down, NULL, NULL, NULL, LLM_FFN_SILU, LLM_FFN_PAR, il);
cb(cur, "ffn_out", il);
cur = ggml_add(ctx0, cur, ffn_inp);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// input for next layer
inpL = cur;
}
cur = inpL;
cur = build_norm(cur, model.output_norm, NULL, LLM_NORM_RMS, -1);
cb(cur, "result_norm", -1);
res->t_embd = cur;
// lm_head
cur = build_lora_mm(model.output, cur);
cb(cur, "result_output", -1);
res->t_logits = cur;
ggml_build_forward_expand(gf, cur);
}
};
struct llm_build_qwen2vl : public llm_graph_context {
llm_build_qwen2vl(const llama_model & model, const llm_graph_params & params, ggml_cgraph * gf) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v;
@ -16487,6 +16606,7 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
case LLM_ARCH_NOMIC_BERT_MOE:
case LLM_ARCH_NEO_BERT:
case LLM_ARCH_WAVTOKENIZER_DEC:
case LLM_ARCH_DREAM:
{
res = nullptr;
} break;
@ -16638,6 +16758,11 @@ llm_graph_result_ptr llama_model::build_graph(
{
llm = std::make_unique<llm_build_qwen2>(*this, params, gf);
} break;
case LLM_ARCH_DREAM:
{
llm = std::make_unique<llm_build_dream>(*this, params, gf);
}
break;
case LLM_ARCH_QWEN2VL:
{
llm = std::make_unique<llm_build_qwen2vl>(*this, params, gf);
@ -17055,6 +17180,7 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
case LLM_ARCH_BITNET:
case LLM_ARCH_QWEN:
case LLM_ARCH_QWEN2:
case LLM_ARCH_DREAM:
case LLM_ARCH_QWEN2MOE:
case LLM_ARCH_QWEN3:
case LLM_ARCH_QWEN3MOE: