Models / qwen3_vl

Qwen3-VL-Embedding-2B: GPU memory to fine-tune

Fine-tuning Qwen3-VL-Embedding-2B at batch 1 and 2048 tokens takes about 37.9 GB of GPU memory for a full fine-tune (fits one A100 / H100 80 GB), and 8.63 GB with LoRA r=16 on a bf16 base (fits one T4 16 GB). Estimated from the model's structure, for the towers the graph reconstructs, not measured. What fits on which card.

Reconstructed from its own config.json with no weights read. 1.4M downloads on Hugging Face.

Our count against the checkpoint

The left number comes from the graph. The right one is the number of scalars in the published weight files. Nothing on this page was tuned to make them agree.

Derived from structure

1.72B

1,723,654,144 parameters

In the published checkpoint

2.13B

2,127,532,032 scalars · safetensors.total, read 2026-09-06

Delta

-19.0%

multi-tower The config declares 2 sub-models (text_config, vision_config). The published checkpoint carries all of them; the graph below carries the towers the reader reconstructs. A gap here is a statement about what we read, not about the checkpoint.

What it costs to run

Cost is a roofline estimate on the priced GPU for 10 epochs at batch 32, 2048 tokens over 50,000 samples (assumed; no dataset attached).

Layers

175

Will it forward-pass

Yes

Priced on

A10G (24GB)

Est. one run

$129.25

Memory: what fits on which card

Largest batch per card. Training is sized at 2048 tokens per sample, serving at 4096 tokens per sequence.

Full fine-tune, batch 1 at 2048 tokens

37.9 GB

LoRA r=16, batch 1 at 2048 tokens

8.63 GB

Serve: bf16 weights

3.45 GB

Serve: KV cache, 4096 tokens

0.47 GB

Card	Full fine-tune	LoRA r=16	Serve bf16, 4096 tokens
T4 16 GB	does not fit at batch 1	batch 2	batch 22
L4 24 GB	2 cards at batch 1 (84 + 91 layers)	batch 3	batch 37
A100 40 GB	2 cards at batch 1 (158 + 17 layers)	batch 6	batch 66
A100 / H100 80 GB	batch 7	batch 13	batch 140
H200 141 GB	batch 18	batch 24	batch 253

Training: 2048 tokens per sample (2048, or the model's context when shorter; the structure table is at the config's maximum, 262144), AdamW with bf16 autocast, SDPA attention (no score matrix stored), 1.15x headroom on every card. Full fine-tune holds weights, gradients and optimizer state (1.2 x 16 bytes per parameter) plus activations. LoRA is r=16 on every linear layer but the head, base frozen in bf16 (2 bytes per parameter), adapters at full AdamW state, same activations. A two-card split is a contiguous layer cut (pipeline parallel), computed only where one card cannot hold batch 1. Serving: bf16 weights plus the KV cache of 4096 tokens per sequence; prefill working set, CUDA context and framework reserve not counted.

These are estimates from the graph, not measurements. Against measured runs, held out by architecture, the estimate flagged 113 of the 131 jobs that did not fit their card and said "fits" on the other 18. Every measured run.

Structure

178 nodes. Output shapes are propagated from the input shape, batch dimension excluded, at 262144 tokens: the config's maximum context, not the 2048 the memory above is sized at.

	Layer	Type	Output shape
1	Input	Input	1 × 262144
2	Embedding	Embedding	1 × 262144 × 2048
3	RoPE	RoPE	1 × 262144 × 2048
4	Vision input	Input	3 × 224 × 224
5	PatchEmbed	Patch Embed	196 × 1024
6	Patch_Position_Embedding	Learned Pos Embed	196 × 1024
7	Vision encoder (internals not in config)	Projection	196 × 2048
8	Vision tokens	Reshape	1 × 196 × 2048
9	Multimodal fusion (concat tokens)	Concatenate	1 × 262340 × 2048
10	RMSNorm_1_1	RMSNorm	1 × 262340 × 2048
11	Attention_1	Grouped Query Attn	1 × 262340 × 2048
12	Add_1_attn	Add	1 × 262340 × 2048
13	RMSNorm_1_2	RMSNorm	1 × 262340 × 2048
14	FFN_1	SwiGLU	1 × 262340 × 2048
15	Add_1_ffn	Add	1 × 262340 × 2048
16	RMSNorm_2_1	RMSNorm	1 × 262340 × 2048
17	Attention_2	Grouped Query Attn	1 × 262340 × 2048
18	Add_2_attn	Add	1 × 262340 × 2048
19	RMSNorm_2_2	RMSNorm	1 × 262340 × 2048
20	FFN_2	SwiGLU	1 × 262340 × 2048
21	Add_2_ffn	Add	1 × 262340 × 2048
22	RMSNorm_3_1	RMSNorm	1 × 262340 × 2048
23	Attention_3	Grouped Query Attn	1 × 262340 × 2048
24	Add_3_attn	Add	1 × 262340 × 2048
25	RMSNorm_3_2	RMSNorm	1 × 262340 × 2048
26	FFN_3	SwiGLU	1 × 262340 × 2048
27	Add_3_ffn	Add	1 × 262340 × 2048
28	RMSNorm_4_1	RMSNorm	1 × 262340 × 2048
29	Attention_4	Grouped Query Attn	1 × 262340 × 2048
30	Add_4_attn	Add	1 × 262340 × 2048
31	RMSNorm_4_2	RMSNorm	1 × 262340 × 2048
32	FFN_4	SwiGLU	1 × 262340 × 2048
33	Add_4_ffn	Add	1 × 262340 × 2048
34	RMSNorm_5_1	RMSNorm	1 × 262340 × 2048
35	Attention_5	Grouped Query Attn	1 × 262340 × 2048
36	Add_5_attn	Add	1 × 262340 × 2048
37	RMSNorm_5_2	RMSNorm	1 × 262340 × 2048
38	FFN_5	SwiGLU	1 × 262340 × 2048
39	Add_5_ffn	Add	1 × 262340 × 2048
40	RMSNorm_6_1	RMSNorm	1 × 262340 × 2048
41	Attention_6	Grouped Query Attn	1 × 262340 × 2048
42	Add_6_attn	Add	1 × 262340 × 2048
43	RMSNorm_6_2	RMSNorm	1 × 262340 × 2048
44	FFN_6	SwiGLU	1 × 262340 × 2048
45	Add_6_ffn	Add	1 × 262340 × 2048
46	RMSNorm_7_1	RMSNorm	1 × 262340 × 2048
47	Attention_7	Grouped Query Attn	1 × 262340 × 2048
48	Add_7_attn	Add	1 × 262340 × 2048
49	RMSNorm_7_2	RMSNorm	1 × 262340 × 2048
50	FFN_7	SwiGLU	1 × 262340 × 2048
51	Add_7_ffn	Add	1 × 262340 × 2048
52	RMSNorm_8_1	RMSNorm	1 × 262340 × 2048
53	Attention_8	Grouped Query Attn	1 × 262340 × 2048
54	Add_8_attn	Add	1 × 262340 × 2048
55	RMSNorm_8_2	RMSNorm	1 × 262340 × 2048
56	FFN_8	SwiGLU	1 × 262340 × 2048
57	Add_8_ffn	Add	1 × 262340 × 2048
58	RMSNorm_9_1	RMSNorm	1 × 262340 × 2048
59	Attention_9	Grouped Query Attn	1 × 262340 × 2048
60	Add_9_attn	Add	1 × 262340 × 2048
61	RMSNorm_9_2	RMSNorm	1 × 262340 × 2048
62	FFN_9	SwiGLU	1 × 262340 × 2048
63	Add_9_ffn	Add	1 × 262340 × 2048
64	RMSNorm_10_1	RMSNorm	1 × 262340 × 2048
65	Attention_10	Grouped Query Attn	1 × 262340 × 2048
66	Add_10_attn	Add	1 × 262340 × 2048
67	RMSNorm_10_2	RMSNorm	1 × 262340 × 2048
68	FFN_10	SwiGLU	1 × 262340 × 2048
69	Add_10_ffn	Add	1 × 262340 × 2048
70	RMSNorm_11_1	RMSNorm	1 × 262340 × 2048
71	Attention_11	Grouped Query Attn	1 × 262340 × 2048
72	Add_11_attn	Add	1 × 262340 × 2048
73	RMSNorm_11_2	RMSNorm	1 × 262340 × 2048
74	FFN_11	SwiGLU	1 × 262340 × 2048
75	Add_11_ffn	Add	1 × 262340 × 2048
76	RMSNorm_12_1	RMSNorm	1 × 262340 × 2048
77	Attention_12	Grouped Query Attn	1 × 262340 × 2048
78	Add_12_attn	Add	1 × 262340 × 2048
79	RMSNorm_12_2	RMSNorm	1 × 262340 × 2048
80	FFN_12	SwiGLU	1 × 262340 × 2048
81	Add_12_ffn	Add	1 × 262340 × 2048
82	RMSNorm_13_1	RMSNorm	1 × 262340 × 2048
83	Attention_13	Grouped Query Attn	1 × 262340 × 2048
84	Add_13_attn	Add	1 × 262340 × 2048
85	RMSNorm_13_2	RMSNorm	1 × 262340 × 2048
86	FFN_13	SwiGLU	1 × 262340 × 2048
87	Add_13_ffn	Add	1 × 262340 × 2048
88	RMSNorm_14_1	RMSNorm	1 × 262340 × 2048
89	Attention_14	Grouped Query Attn	1 × 262340 × 2048
90	Add_14_attn	Add	1 × 262340 × 2048
91	RMSNorm_14_2	RMSNorm	1 × 262340 × 2048
92	FFN_14	SwiGLU	1 × 262340 × 2048
93	Add_14_ffn	Add	1 × 262340 × 2048
94	RMSNorm_15_1	RMSNorm	1 × 262340 × 2048
95	Attention_15	Grouped Query Attn	1 × 262340 × 2048
96	Add_15_attn	Add	1 × 262340 × 2048
97	RMSNorm_15_2	RMSNorm	1 × 262340 × 2048
98	FFN_15	SwiGLU	1 × 262340 × 2048
99	Add_15_ffn	Add	1 × 262340 × 2048
100	RMSNorm_16_1	RMSNorm	1 × 262340 × 2048
101	Attention_16	Grouped Query Attn	1 × 262340 × 2048
102	Add_16_attn	Add	1 × 262340 × 2048
103	RMSNorm_16_2	RMSNorm	1 × 262340 × 2048
104	FFN_16	SwiGLU	1 × 262340 × 2048
105	Add_16_ffn	Add	1 × 262340 × 2048
106	RMSNorm_17_1	RMSNorm	1 × 262340 × 2048
107	Attention_17	Grouped Query Attn	1 × 262340 × 2048
108	Add_17_attn	Add	1 × 262340 × 2048
109	RMSNorm_17_2	RMSNorm	1 × 262340 × 2048
110	FFN_17	SwiGLU	1 × 262340 × 2048
111	Add_17_ffn	Add	1 × 262340 × 2048
112	RMSNorm_18_1	RMSNorm	1 × 262340 × 2048
113	Attention_18	Grouped Query Attn	1 × 262340 × 2048
114	Add_18_attn	Add	1 × 262340 × 2048
115	RMSNorm_18_2	RMSNorm	1 × 262340 × 2048
116	FFN_18	SwiGLU	1 × 262340 × 2048
117	Add_18_ffn	Add	1 × 262340 × 2048
118	RMSNorm_19_1	RMSNorm	1 × 262340 × 2048
119	Attention_19	Grouped Query Attn	1 × 262340 × 2048
120	Add_19_attn	Add	1 × 262340 × 2048
121	RMSNorm_19_2	RMSNorm	1 × 262340 × 2048
122	FFN_19	SwiGLU	1 × 262340 × 2048
123	Add_19_ffn	Add	1 × 262340 × 2048
124	RMSNorm_20_1	RMSNorm	1 × 262340 × 2048
125	Attention_20	Grouped Query Attn	1 × 262340 × 2048
126	Add_20_attn	Add	1 × 262340 × 2048
127	RMSNorm_20_2	RMSNorm	1 × 262340 × 2048
128	FFN_20	SwiGLU	1 × 262340 × 2048
129	Add_20_ffn	Add	1 × 262340 × 2048
130	RMSNorm_21_1	RMSNorm	1 × 262340 × 2048
131	Attention_21	Grouped Query Attn	1 × 262340 × 2048
132	Add_21_attn	Add	1 × 262340 × 2048
133	RMSNorm_21_2	RMSNorm	1 × 262340 × 2048
134	FFN_21	SwiGLU	1 × 262340 × 2048
135	Add_21_ffn	Add	1 × 262340 × 2048
136	RMSNorm_22_1	RMSNorm	1 × 262340 × 2048
137	Attention_22	Grouped Query Attn	1 × 262340 × 2048
138	Add_22_attn	Add	1 × 262340 × 2048
139	RMSNorm_22_2	RMSNorm	1 × 262340 × 2048
140	FFN_22	SwiGLU	1 × 262340 × 2048
141	Add_22_ffn	Add	1 × 262340 × 2048
142	RMSNorm_23_1	RMSNorm	1 × 262340 × 2048
143	Attention_23	Grouped Query Attn	1 × 262340 × 2048
144	Add_23_attn	Add	1 × 262340 × 2048
145	RMSNorm_23_2	RMSNorm	1 × 262340 × 2048
146	FFN_23	SwiGLU	1 × 262340 × 2048
147	Add_23_ffn	Add	1 × 262340 × 2048
148	RMSNorm_24_1	RMSNorm	1 × 262340 × 2048
149	Attention_24	Grouped Query Attn	1 × 262340 × 2048
150	Add_24_attn	Add	1 × 262340 × 2048
151	RMSNorm_24_2	RMSNorm	1 × 262340 × 2048
152	FFN_24	SwiGLU	1 × 262340 × 2048
153	Add_24_ffn	Add	1 × 262340 × 2048
154	RMSNorm_25_1	RMSNorm	1 × 262340 × 2048
155	Attention_25	Grouped Query Attn	1 × 262340 × 2048
156	Add_25_attn	Add	1 × 262340 × 2048
157	RMSNorm_25_2	RMSNorm	1 × 262340 × 2048
158	FFN_25	SwiGLU	1 × 262340 × 2048
159	Add_25_ffn	Add	1 × 262340 × 2048
160	RMSNorm_26_1	RMSNorm	1 × 262340 × 2048
161	Attention_26	Grouped Query Attn	1 × 262340 × 2048
162	Add_26_attn	Add	1 × 262340 × 2048
163	RMSNorm_26_2	RMSNorm	1 × 262340 × 2048
164	FFN_26	SwiGLU	1 × 262340 × 2048
165	Add_26_ffn	Add	1 × 262340 × 2048
166	RMSNorm_27_1	RMSNorm	1 × 262340 × 2048
167	Attention_27	Grouped Query Attn	1 × 262340 × 2048
168	Add_27_attn	Add	1 × 262340 × 2048
169	RMSNorm_27_2	RMSNorm	1 × 262340 × 2048
170	FFN_27	SwiGLU	1 × 262340 × 2048
171	Add_27_ffn	Add	1 × 262340 × 2048
172	RMSNorm_28_1	RMSNorm	1 × 262340 × 2048
173	Attention_28	Grouped Query Attn	1 × 262340 × 2048
174	Add_28_attn	Add	1 × 262340 × 2048
175	RMSNorm_28_2	RMSNorm	1 × 262340 × 2048
176	FFN_28	SwiGLU	1 × 262340 × 2048
177	Add_28_ffn	Add	1 × 262340 × 2048
178	Output	Output	1 × 262340 × 2048

What the verifier says

warn"RoPE" receives input but its output is not connected. This layer will be unreachable in the forward pass. Fix: Connect the output forward, or add an Output node if this is the final layer.
dead-end

infoAt 28 stacked attention layers, residual-branch outputs add up; unscaled init lets activation variance grow with depth. GPT-2/LLaMA-family models scale the residual projections by depth (N(0, 0.02 / √(2L))). Fix: Scale residual output projections by depth: nn.init.normal_(w, std=0.02 / math.sqrt(2 * n_layers))
deep-attention-default-init

What changed upstream

The change tracker follows Qwen/Qwen3-VL-Embedding-2B: every revision it has read, what changed in the config, tokenizer, chat template or weights between them, and an OSV advisory per change.

Revision history and changes Badge JSON OSV advisories All tracked changes

Do this to your own model

Same numbers, on a model in your repo, in one command. No account.

pip install neurarch-trace
neurarch-trace Qwen/Qwen3-VL-Embedding-2B --plan --share

Paste a model instead Markdown twin Index JSON POST /api/v1/plan