Models / gemma4

OTel-2.0-LLM-31B-IT

Reconstructed from its own config.json with no weights read. 6.8M 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

30.56B

30,564,872,304 parameters

In the published checkpoint

31.27B

31,273,088,876 scalars · safetensors.total, read 2026-09-06

Delta

-2.26%

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 over 50,000 samples (assumed; no dataset attached). GPU fit is fp32 weights plus gradients plus two Adam moments (16 bytes per parameter) with 1.3x headroom; activations are not included and grow with batch size.

Layers

475

Will it forward-pass

Yes

Priced on

A10G (24GB)

Est. one run

$1548238.17

Card	Memory
T4 (16GB)	weights + activations	does not fit
A100 (40GB)	weights + activations	does not fit
H100 (80GB)	weights + activations	does not fit

Structure

478 nodes. Output shapes are propagated from the input shape, batch dimension excluded.

	Layer	Type	Output shape
1	Input	Input	1 × 262144
2	Embedding	Embedding	1 × 262144 × 5376
3	RoPE	RoPE	1 × 262144 × 5376
4	Vision input	Input	3 × 224 × 224
5	PatchEmbed	Patch Embed	196 × 1152
6	Patch_Position_Embedding	Learned Pos Embed	196 × 1152
7	Vision_LN_1	LayerNorm	196 × 1152
8	Vision_Attn_1	Multi-Head Attention	196 × 1152
9	Vision_Add_1	Add	196 × 1152
10	Vision_FFN_1	Feed Forward	196 × 1152
11	Vision_LN_2	LayerNorm	196 × 1152
12	Vision_Attn_2	Multi-Head Attention	196 × 1152
13	Vision_Add_2	Add	196 × 1152
14	Vision_FFN_2	Feed Forward	196 × 1152
15	Vision_LN_3	LayerNorm	196 × 1152
16	Vision_Attn_3	Multi-Head Attention	196 × 1152
17	Vision_Add_3	Add	196 × 1152
18	Vision_FFN_3	Feed Forward	196 × 1152
19	Vision_LN_4	LayerNorm	196 × 1152
20	Vision_Attn_4	Multi-Head Attention	196 × 1152
21	Vision_Add_4	Add	196 × 1152
22	Vision_FFN_4	Feed Forward	196 × 1152
23	Vision_LN_5	LayerNorm	196 × 1152
24	Vision_Attn_5	Multi-Head Attention	196 × 1152
25	Vision_Add_5	Add	196 × 1152
26	Vision_FFN_5	Feed Forward	196 × 1152
27	Vision_LN_6	LayerNorm	196 × 1152
28	Vision_Attn_6	Multi-Head Attention	196 × 1152
29	Vision_Add_6	Add	196 × 1152
30	Vision_FFN_6	Feed Forward	196 × 1152
31	Vision_LN_7	LayerNorm	196 × 1152
32	Vision_Attn_7	Multi-Head Attention	196 × 1152
33	Vision_Add_7	Add	196 × 1152
34	Vision_FFN_7	Feed Forward	196 × 1152
35	Vision_LN_8	LayerNorm	196 × 1152
36	Vision_Attn_8	Multi-Head Attention	196 × 1152
37	Vision_Add_8	Add	196 × 1152
38	Vision_FFN_8	Feed Forward	196 × 1152
39	Vision_LN_9	LayerNorm	196 × 1152
40	Vision_Attn_9	Multi-Head Attention	196 × 1152
41	Vision_Add_9	Add	196 × 1152
42	Vision_FFN_9	Feed Forward	196 × 1152
43	Vision_LN_10	LayerNorm	196 × 1152
44	Vision_Attn_10	Multi-Head Attention	196 × 1152
45	Vision_Add_10	Add	196 × 1152
46	Vision_FFN_10	Feed Forward	196 × 1152
47	Vision_LN_11	LayerNorm	196 × 1152
48	Vision_Attn_11	Multi-Head Attention	196 × 1152
49	Vision_Add_11	Add	196 × 1152
50	Vision_FFN_11	Feed Forward	196 × 1152
51	Vision_LN_12	LayerNorm	196 × 1152
52	Vision_Attn_12	Multi-Head Attention	196 × 1152
53	Vision_Add_12	Add	196 × 1152
54	Vision_FFN_12	Feed Forward	196 × 1152
55	Vision_LN_13	LayerNorm	196 × 1152
56	Vision_Attn_13	Multi-Head Attention	196 × 1152
57	Vision_Add_13	Add	196 × 1152
58	Vision_FFN_13	Feed Forward	196 × 1152
59	Vision_LN_14	LayerNorm	196 × 1152
60	Vision_Attn_14	Multi-Head Attention	196 × 1152
61	Vision_Add_14	Add	196 × 1152
62	Vision_FFN_14	Feed Forward	196 × 1152
63	Vision_LN_15	LayerNorm	196 × 1152
64	Vision_Attn_15	Multi-Head Attention	196 × 1152
65	Vision_Add_15	Add	196 × 1152
66	Vision_FFN_15	Feed Forward	196 × 1152
67	Vision_LN_16	LayerNorm	196 × 1152
68	Vision_Attn_16	Multi-Head Attention	196 × 1152
69	Vision_Add_16	Add	196 × 1152
70	Vision_FFN_16	Feed Forward	196 × 1152
71	Vision_LN_17	LayerNorm	196 × 1152
72	Vision_Attn_17	Multi-Head Attention	196 × 1152
73	Vision_Add_17	Add	196 × 1152
74	Vision_FFN_17	Feed Forward	196 × 1152
75	Vision_LN_18	LayerNorm	196 × 1152
76	Vision_Attn_18	Multi-Head Attention	196 × 1152
77	Vision_Add_18	Add	196 × 1152
78	Vision_FFN_18	Feed Forward	196 × 1152
79	Vision_LN_19	LayerNorm	196 × 1152
80	Vision_Attn_19	Multi-Head Attention	196 × 1152
81	Vision_Add_19	Add	196 × 1152
82	Vision_FFN_19	Feed Forward	196 × 1152
83	Vision_LN_20	LayerNorm	196 × 1152
84	Vision_Attn_20	Multi-Head Attention	196 × 1152
85	Vision_Add_20	Add	196 × 1152
86	Vision_FFN_20	Feed Forward	196 × 1152
87	Vision_LN_21	LayerNorm	196 × 1152
88	Vision_Attn_21	Multi-Head Attention	196 × 1152
89	Vision_Add_21	Add	196 × 1152
90	Vision_FFN_21	Feed Forward	196 × 1152
91	Vision_LN_22	LayerNorm	196 × 1152
92	Vision_Attn_22	Multi-Head Attention	196 × 1152
93	Vision_Add_22	Add	196 × 1152
94	Vision_FFN_22	Feed Forward	196 × 1152
95	Vision_LN_23	LayerNorm	196 × 1152
96	Vision_Attn_23	Multi-Head Attention	196 × 1152
97	Vision_Add_23	Add	196 × 1152
98	Vision_FFN_23	Feed Forward	196 × 1152
99	Vision_LN_24	LayerNorm	196 × 1152
100	Vision_Attn_24	Multi-Head Attention	196 × 1152
101	Vision_Add_24	Add	196 × 1152
102	Vision_FFN_24	Feed Forward	196 × 1152
103	Vision_LN_25	LayerNorm	196 × 1152
104	Vision_Attn_25	Multi-Head Attention	196 × 1152
105	Vision_Add_25	Add	196 × 1152
106	Vision_FFN_25	Feed Forward	196 × 1152
107	Vision_LN_26	LayerNorm	196 × 1152
108	Vision_Attn_26	Multi-Head Attention	196 × 1152
109	Vision_Add_26	Add	196 × 1152
110	Vision_FFN_26	Feed Forward	196 × 1152
111	Vision_LN_27	LayerNorm	196 × 1152
112	Vision_Attn_27	Multi-Head Attention	196 × 1152
113	Vision_Add_27	Add	196 × 1152
114	Vision_FFN_27	Feed Forward	196 × 1152
115	Vision projector	Projection	196 × 5376
116	Vision tokens	Reshape	1 × 196 × 5376
117	Multimodal fusion (concat tokens)	Concatenate	1 × 262340 × 5376
118	RMSNorm_1_1	RMSNorm	1 × 262340 × 5376
119	Attention_1	Grouped Query Attn	1 × 262340 × 5376
120	Add_1_attn	Add	1 × 262340 × 5376
121	RMSNorm_1_2	RMSNorm	1 × 262340 × 5376
122	FFN_1	SwiGLU	1 × 262340 × 5376
123	Add_1_ffn	Add	1 × 262340 × 5376
124	RMSNorm_2_1	RMSNorm	1 × 262340 × 5376
125	Attention_2	Grouped Query Attn	1 × 262340 × 5376
126	Add_2_attn	Add	1 × 262340 × 5376
127	RMSNorm_2_2	RMSNorm	1 × 262340 × 5376
128	FFN_2	SwiGLU	1 × 262340 × 5376
129	Add_2_ffn	Add	1 × 262340 × 5376
130	RMSNorm_3_1	RMSNorm	1 × 262340 × 5376
131	Attention_3	Grouped Query Attn	1 × 262340 × 5376
132	Add_3_attn	Add	1 × 262340 × 5376
133	RMSNorm_3_2	RMSNorm	1 × 262340 × 5376
134	FFN_3	SwiGLU	1 × 262340 × 5376
135	Add_3_ffn	Add	1 × 262340 × 5376
136	RMSNorm_4_1	RMSNorm	1 × 262340 × 5376
137	Attention_4	Grouped Query Attn	1 × 262340 × 5376
138	Add_4_attn	Add	1 × 262340 × 5376
139	RMSNorm_4_2	RMSNorm	1 × 262340 × 5376
140	FFN_4	SwiGLU	1 × 262340 × 5376
141	Add_4_ffn	Add	1 × 262340 × 5376
142	RMSNorm_5_1	RMSNorm	1 × 262340 × 5376
143	Attention_5	Grouped Query Attn	1 × 262340 × 5376
144	Add_5_attn	Add	1 × 262340 × 5376
145	RMSNorm_5_2	RMSNorm	1 × 262340 × 5376
146	FFN_5	SwiGLU	1 × 262340 × 5376
147	Add_5_ffn	Add	1 × 262340 × 5376
148	RMSNorm_6_1	RMSNorm	1 × 262340 × 5376
149	Attention_6	Grouped Query Attn	1 × 262340 × 5376
150	Add_6_attn	Add	1 × 262340 × 5376
151	RMSNorm_6_2	RMSNorm	1 × 262340 × 5376
152	FFN_6	SwiGLU	1 × 262340 × 5376
153	Add_6_ffn	Add	1 × 262340 × 5376
154	RMSNorm_7_1	RMSNorm	1 × 262340 × 5376
155	Attention_7	Grouped Query Attn	1 × 262340 × 5376
156	Add_7_attn	Add	1 × 262340 × 5376
157	RMSNorm_7_2	RMSNorm	1 × 262340 × 5376
158	FFN_7	SwiGLU	1 × 262340 × 5376
159	Add_7_ffn	Add	1 × 262340 × 5376
160	RMSNorm_8_1	RMSNorm	1 × 262340 × 5376
161	Attention_8	Grouped Query Attn	1 × 262340 × 5376
162	Add_8_attn	Add	1 × 262340 × 5376
163	RMSNorm_8_2	RMSNorm	1 × 262340 × 5376
164	FFN_8	SwiGLU	1 × 262340 × 5376
165	Add_8_ffn	Add	1 × 262340 × 5376
166	RMSNorm_9_1	RMSNorm	1 × 262340 × 5376
167	Attention_9	Grouped Query Attn	1 × 262340 × 5376
168	Add_9_attn	Add	1 × 262340 × 5376
169	RMSNorm_9_2	RMSNorm	1 × 262340 × 5376
170	FFN_9	SwiGLU	1 × 262340 × 5376
171	Add_9_ffn	Add	1 × 262340 × 5376
172	RMSNorm_10_1	RMSNorm	1 × 262340 × 5376
173	Attention_10	Grouped Query Attn	1 × 262340 × 5376
174	Add_10_attn	Add	1 × 262340 × 5376
175	RMSNorm_10_2	RMSNorm	1 × 262340 × 5376
176	FFN_10	SwiGLU	1 × 262340 × 5376
177	Add_10_ffn	Add	1 × 262340 × 5376
178	RMSNorm_11_1	RMSNorm	1 × 262340 × 5376
179	Attention_11	Grouped Query Attn	1 × 262340 × 5376
180	Add_11_attn	Add	1 × 262340 × 5376
181	RMSNorm_11_2	RMSNorm	1 × 262340 × 5376
182	FFN_11	SwiGLU	1 × 262340 × 5376
183	Add_11_ffn	Add	1 × 262340 × 5376
184	RMSNorm_12_1	RMSNorm	1 × 262340 × 5376
185	Attention_12	Grouped Query Attn	1 × 262340 × 5376
186	Add_12_attn	Add	1 × 262340 × 5376
187	RMSNorm_12_2	RMSNorm	1 × 262340 × 5376
188	FFN_12	SwiGLU	1 × 262340 × 5376
189	Add_12_ffn	Add	1 × 262340 × 5376
190	RMSNorm_13_1	RMSNorm	1 × 262340 × 5376
191	Attention_13	Grouped Query Attn	1 × 262340 × 5376
192	Add_13_attn	Add	1 × 262340 × 5376
193	RMSNorm_13_2	RMSNorm	1 × 262340 × 5376
194	FFN_13	SwiGLU	1 × 262340 × 5376
195	Add_13_ffn	Add	1 × 262340 × 5376
196	RMSNorm_14_1	RMSNorm	1 × 262340 × 5376
197	Attention_14	Grouped Query Attn	1 × 262340 × 5376
198	Add_14_attn	Add	1 × 262340 × 5376
199	RMSNorm_14_2	RMSNorm	1 × 262340 × 5376
200	FFN_14	SwiGLU	1 × 262340 × 5376
201	Add_14_ffn	Add	1 × 262340 × 5376
202	RMSNorm_15_1	RMSNorm	1 × 262340 × 5376
203	Attention_15	Grouped Query Attn	1 × 262340 × 5376
204	Add_15_attn	Add	1 × 262340 × 5376
205	RMSNorm_15_2	RMSNorm	1 × 262340 × 5376
206	FFN_15	SwiGLU	1 × 262340 × 5376
207	Add_15_ffn	Add	1 × 262340 × 5376
208	RMSNorm_16_1	RMSNorm	1 × 262340 × 5376
209	Attention_16	Grouped Query Attn	1 × 262340 × 5376
210	Add_16_attn	Add	1 × 262340 × 5376
211	RMSNorm_16_2	RMSNorm	1 × 262340 × 5376
212	FFN_16	SwiGLU	1 × 262340 × 5376
213	Add_16_ffn	Add	1 × 262340 × 5376
214	RMSNorm_17_1	RMSNorm	1 × 262340 × 5376
215	Attention_17	Grouped Query Attn	1 × 262340 × 5376
216	Add_17_attn	Add	1 × 262340 × 5376
217	RMSNorm_17_2	RMSNorm	1 × 262340 × 5376
218	FFN_17	SwiGLU	1 × 262340 × 5376
219	Add_17_ffn	Add	1 × 262340 × 5376
220	RMSNorm_18_1	RMSNorm	1 × 262340 × 5376
221	Attention_18	Grouped Query Attn	1 × 262340 × 5376
222	Add_18_attn	Add	1 × 262340 × 5376
223	RMSNorm_18_2	RMSNorm	1 × 262340 × 5376
224	FFN_18	SwiGLU	1 × 262340 × 5376
225	Add_18_ffn	Add	1 × 262340 × 5376
226	RMSNorm_19_1	RMSNorm	1 × 262340 × 5376
227	Attention_19	Grouped Query Attn	1 × 262340 × 5376
228	Add_19_attn	Add	1 × 262340 × 5376
229	RMSNorm_19_2	RMSNorm	1 × 262340 × 5376
230	FFN_19	SwiGLU	1 × 262340 × 5376
231	Add_19_ffn	Add	1 × 262340 × 5376
232	RMSNorm_20_1	RMSNorm	1 × 262340 × 5376
233	Attention_20	Grouped Query Attn	1 × 262340 × 5376
234	Add_20_attn	Add	1 × 262340 × 5376
235	RMSNorm_20_2	RMSNorm	1 × 262340 × 5376
236	FFN_20	SwiGLU	1 × 262340 × 5376
237	Add_20_ffn	Add	1 × 262340 × 5376
238	RMSNorm_21_1	RMSNorm	1 × 262340 × 5376
239	Attention_21	Grouped Query Attn	1 × 262340 × 5376
240	Add_21_attn	Add	1 × 262340 × 5376
241	RMSNorm_21_2	RMSNorm	1 × 262340 × 5376
242	FFN_21	SwiGLU	1 × 262340 × 5376
243	Add_21_ffn	Add	1 × 262340 × 5376
244	RMSNorm_22_1	RMSNorm	1 × 262340 × 5376
245	Attention_22	Grouped Query Attn	1 × 262340 × 5376
246	Add_22_attn	Add	1 × 262340 × 5376
247	RMSNorm_22_2	RMSNorm	1 × 262340 × 5376
248	FFN_22	SwiGLU	1 × 262340 × 5376
249	Add_22_ffn	Add	1 × 262340 × 5376
250	RMSNorm_23_1	RMSNorm	1 × 262340 × 5376
251	Attention_23	Grouped Query Attn	1 × 262340 × 5376
252	Add_23_attn	Add	1 × 262340 × 5376
253	RMSNorm_23_2	RMSNorm	1 × 262340 × 5376
254	FFN_23	SwiGLU	1 × 262340 × 5376
255	Add_23_ffn	Add	1 × 262340 × 5376
256	RMSNorm_24_1	RMSNorm	1 × 262340 × 5376
257	Attention_24	Grouped Query Attn	1 × 262340 × 5376
258	Add_24_attn	Add	1 × 262340 × 5376
259	RMSNorm_24_2	RMSNorm	1 × 262340 × 5376
260	FFN_24	SwiGLU	1 × 262340 × 5376
261	Add_24_ffn	Add	1 × 262340 × 5376
262	RMSNorm_25_1	RMSNorm	1 × 262340 × 5376
263	Attention_25	Grouped Query Attn	1 × 262340 × 5376
264	Add_25_attn	Add	1 × 262340 × 5376
265	RMSNorm_25_2	RMSNorm	1 × 262340 × 5376
266	FFN_25	SwiGLU	1 × 262340 × 5376
267	Add_25_ffn	Add	1 × 262340 × 5376
268	RMSNorm_26_1	RMSNorm	1 × 262340 × 5376
269	Attention_26	Grouped Query Attn	1 × 262340 × 5376
270	Add_26_attn	Add	1 × 262340 × 5376
271	RMSNorm_26_2	RMSNorm	1 × 262340 × 5376
272	FFN_26	SwiGLU	1 × 262340 × 5376
273	Add_26_ffn	Add	1 × 262340 × 5376
274	RMSNorm_27_1	RMSNorm	1 × 262340 × 5376
275	Attention_27	Grouped Query Attn	1 × 262340 × 5376
276	Add_27_attn	Add	1 × 262340 × 5376
277	RMSNorm_27_2	RMSNorm	1 × 262340 × 5376
278	FFN_27	SwiGLU	1 × 262340 × 5376
279	Add_27_ffn	Add	1 × 262340 × 5376
280	RMSNorm_28_1	RMSNorm	1 × 262340 × 5376
281	Attention_28	Grouped Query Attn	1 × 262340 × 5376
282	Add_28_attn	Add	1 × 262340 × 5376
283	RMSNorm_28_2	RMSNorm	1 × 262340 × 5376
284	FFN_28	SwiGLU	1 × 262340 × 5376
285	Add_28_ffn	Add	1 × 262340 × 5376
286	RMSNorm_29_1	RMSNorm	1 × 262340 × 5376
287	Attention_29	Grouped Query Attn	1 × 262340 × 5376
288	Add_29_attn	Add	1 × 262340 × 5376
289	RMSNorm_29_2	RMSNorm	1 × 262340 × 5376
290	FFN_29	SwiGLU	1 × 262340 × 5376
291	Add_29_ffn	Add	1 × 262340 × 5376
292	RMSNorm_30_1	RMSNorm	1 × 262340 × 5376
293	Attention_30	Grouped Query Attn	1 × 262340 × 5376
294	Add_30_attn	Add	1 × 262340 × 5376
295	RMSNorm_30_2	RMSNorm	1 × 262340 × 5376
296	FFN_30	SwiGLU	1 × 262340 × 5376
297	Add_30_ffn	Add	1 × 262340 × 5376
298	RMSNorm_31_1	RMSNorm	1 × 262340 × 5376
299	Attention_31	Grouped Query Attn	1 × 262340 × 5376
300	Add_31_attn	Add	1 × 262340 × 5376
301	RMSNorm_31_2	RMSNorm	1 × 262340 × 5376
302	FFN_31	SwiGLU	1 × 262340 × 5376
303	Add_31_ffn	Add	1 × 262340 × 5376
304	RMSNorm_32_1	RMSNorm	1 × 262340 × 5376
305	Attention_32	Grouped Query Attn	1 × 262340 × 5376
306	Add_32_attn	Add	1 × 262340 × 5376
307	RMSNorm_32_2	RMSNorm	1 × 262340 × 5376
308	FFN_32	SwiGLU	1 × 262340 × 5376
309	Add_32_ffn	Add	1 × 262340 × 5376
310	RMSNorm_33_1	RMSNorm	1 × 262340 × 5376
311	Attention_33	Grouped Query Attn	1 × 262340 × 5376
312	Add_33_attn	Add	1 × 262340 × 5376
313	RMSNorm_33_2	RMSNorm	1 × 262340 × 5376
314	FFN_33	SwiGLU	1 × 262340 × 5376
315	Add_33_ffn	Add	1 × 262340 × 5376
316	RMSNorm_34_1	RMSNorm	1 × 262340 × 5376
317	Attention_34	Grouped Query Attn	1 × 262340 × 5376
318	Add_34_attn	Add	1 × 262340 × 5376
319	RMSNorm_34_2	RMSNorm	1 × 262340 × 5376
320	FFN_34	SwiGLU	1 × 262340 × 5376
321	Add_34_ffn	Add	1 × 262340 × 5376
322	RMSNorm_35_1	RMSNorm	1 × 262340 × 5376
323	Attention_35	Grouped Query Attn	1 × 262340 × 5376
324	Add_35_attn	Add	1 × 262340 × 5376
325	RMSNorm_35_2	RMSNorm	1 × 262340 × 5376
326	FFN_35	SwiGLU	1 × 262340 × 5376
327	Add_35_ffn	Add	1 × 262340 × 5376
328	RMSNorm_36_1	RMSNorm	1 × 262340 × 5376
329	Attention_36	Grouped Query Attn	1 × 262340 × 5376
330	Add_36_attn	Add	1 × 262340 × 5376
331	RMSNorm_36_2	RMSNorm	1 × 262340 × 5376
332	FFN_36	SwiGLU	1 × 262340 × 5376
333	Add_36_ffn	Add	1 × 262340 × 5376
334	RMSNorm_37_1	RMSNorm	1 × 262340 × 5376
335	Attention_37	Grouped Query Attn	1 × 262340 × 5376
336	Add_37_attn	Add	1 × 262340 × 5376
337	RMSNorm_37_2	RMSNorm	1 × 262340 × 5376
338	FFN_37	SwiGLU	1 × 262340 × 5376
339	Add_37_ffn	Add	1 × 262340 × 5376
340	RMSNorm_38_1	RMSNorm	1 × 262340 × 5376
341	Attention_38	Grouped Query Attn	1 × 262340 × 5376
342	Add_38_attn	Add	1 × 262340 × 5376
343	RMSNorm_38_2	RMSNorm	1 × 262340 × 5376
344	FFN_38	SwiGLU	1 × 262340 × 5376
345	Add_38_ffn	Add	1 × 262340 × 5376
346	RMSNorm_39_1	RMSNorm	1 × 262340 × 5376
347	Attention_39	Grouped Query Attn	1 × 262340 × 5376
348	Add_39_attn	Add	1 × 262340 × 5376
349	RMSNorm_39_2	RMSNorm	1 × 262340 × 5376
350	FFN_39	SwiGLU	1 × 262340 × 5376
351	Add_39_ffn	Add	1 × 262340 × 5376
352	RMSNorm_40_1	RMSNorm	1 × 262340 × 5376
353	Attention_40	Grouped Query Attn	1 × 262340 × 5376
354	Add_40_attn	Add	1 × 262340 × 5376
355	RMSNorm_40_2	RMSNorm	1 × 262340 × 5376
356	FFN_40	SwiGLU	1 × 262340 × 5376
357	Add_40_ffn	Add	1 × 262340 × 5376
358	RMSNorm_41_1	RMSNorm	1 × 262340 × 5376
359	Attention_41	Grouped Query Attn	1 × 262340 × 5376
360	Add_41_attn	Add	1 × 262340 × 5376
361	RMSNorm_41_2	RMSNorm	1 × 262340 × 5376
362	FFN_41	SwiGLU	1 × 262340 × 5376
363	Add_41_ffn	Add	1 × 262340 × 5376
364	RMSNorm_42_1	RMSNorm	1 × 262340 × 5376
365	Attention_42	Grouped Query Attn	1 × 262340 × 5376
366	Add_42_attn	Add	1 × 262340 × 5376
367	RMSNorm_42_2	RMSNorm	1 × 262340 × 5376
368	FFN_42	SwiGLU	1 × 262340 × 5376
369	Add_42_ffn	Add	1 × 262340 × 5376
370	RMSNorm_43_1	RMSNorm	1 × 262340 × 5376
371	Attention_43	Grouped Query Attn	1 × 262340 × 5376
372	Add_43_attn	Add	1 × 262340 × 5376
373	RMSNorm_43_2	RMSNorm	1 × 262340 × 5376
374	FFN_43	SwiGLU	1 × 262340 × 5376
375	Add_43_ffn	Add	1 × 262340 × 5376
376	RMSNorm_44_1	RMSNorm	1 × 262340 × 5376
377	Attention_44	Grouped Query Attn	1 × 262340 × 5376
378	Add_44_attn	Add	1 × 262340 × 5376
379	RMSNorm_44_2	RMSNorm	1 × 262340 × 5376
380	FFN_44	SwiGLU	1 × 262340 × 5376
381	Add_44_ffn	Add	1 × 262340 × 5376
382	RMSNorm_45_1	RMSNorm	1 × 262340 × 5376
383	Attention_45	Grouped Query Attn	1 × 262340 × 5376
384	Add_45_attn	Add	1 × 262340 × 5376
385	RMSNorm_45_2	RMSNorm	1 × 262340 × 5376
386	FFN_45	SwiGLU	1 × 262340 × 5376
387	Add_45_ffn	Add	1 × 262340 × 5376
388	RMSNorm_46_1	RMSNorm	1 × 262340 × 5376
389	Attention_46	Grouped Query Attn	1 × 262340 × 5376
390	Add_46_attn	Add	1 × 262340 × 5376
391	RMSNorm_46_2	RMSNorm	1 × 262340 × 5376
392	FFN_46	SwiGLU	1 × 262340 × 5376
393	Add_46_ffn	Add	1 × 262340 × 5376
394	RMSNorm_47_1	RMSNorm	1 × 262340 × 5376
395	Attention_47	Grouped Query Attn	1 × 262340 × 5376
396	Add_47_attn	Add	1 × 262340 × 5376
397	RMSNorm_47_2	RMSNorm	1 × 262340 × 5376
398	FFN_47	SwiGLU	1 × 262340 × 5376
399	Add_47_ffn	Add	1 × 262340 × 5376
400	RMSNorm_48_1	RMSNorm	1 × 262340 × 5376
401	Attention_48	Grouped Query Attn	1 × 262340 × 5376
402	Add_48_attn	Add	1 × 262340 × 5376
403	RMSNorm_48_2	RMSNorm	1 × 262340 × 5376
404	FFN_48	SwiGLU	1 × 262340 × 5376
405	Add_48_ffn	Add	1 × 262340 × 5376
406	RMSNorm_49_1	RMSNorm	1 × 262340 × 5376
407	Attention_49	Grouped Query Attn	1 × 262340 × 5376
408	Add_49_attn	Add	1 × 262340 × 5376
409	RMSNorm_49_2	RMSNorm	1 × 262340 × 5376
410	FFN_49	SwiGLU	1 × 262340 × 5376
411	Add_49_ffn	Add	1 × 262340 × 5376
412	RMSNorm_50_1	RMSNorm	1 × 262340 × 5376
413	Attention_50	Grouped Query Attn	1 × 262340 × 5376
414	Add_50_attn	Add	1 × 262340 × 5376
415	RMSNorm_50_2	RMSNorm	1 × 262340 × 5376
416	FFN_50	SwiGLU	1 × 262340 × 5376
417	Add_50_ffn	Add	1 × 262340 × 5376
418	RMSNorm_51_1	RMSNorm	1 × 262340 × 5376
419	Attention_51	Grouped Query Attn	1 × 262340 × 5376
420	Add_51_attn	Add	1 × 262340 × 5376
421	RMSNorm_51_2	RMSNorm	1 × 262340 × 5376
422	FFN_51	SwiGLU	1 × 262340 × 5376
423	Add_51_ffn	Add	1 × 262340 × 5376
424	RMSNorm_52_1	RMSNorm	1 × 262340 × 5376
425	Attention_52	Grouped Query Attn	1 × 262340 × 5376
426	Add_52_attn	Add	1 × 262340 × 5376
427	RMSNorm_52_2	RMSNorm	1 × 262340 × 5376
428	FFN_52	SwiGLU	1 × 262340 × 5376
429	Add_52_ffn	Add	1 × 262340 × 5376
430	RMSNorm_53_1	RMSNorm	1 × 262340 × 5376
431	Attention_53	Grouped Query Attn	1 × 262340 × 5376
432	Add_53_attn	Add	1 × 262340 × 5376
433	RMSNorm_53_2	RMSNorm	1 × 262340 × 5376
434	FFN_53	SwiGLU	1 × 262340 × 5376
435	Add_53_ffn	Add	1 × 262340 × 5376
436	RMSNorm_54_1	RMSNorm	1 × 262340 × 5376
437	Attention_54	Grouped Query Attn	1 × 262340 × 5376
438	Add_54_attn	Add	1 × 262340 × 5376
439	RMSNorm_54_2	RMSNorm	1 × 262340 × 5376
440	FFN_54	SwiGLU	1 × 262340 × 5376
441	Add_54_ffn	Add	1 × 262340 × 5376
442	RMSNorm_55_1	RMSNorm	1 × 262340 × 5376
443	Attention_55	Grouped Query Attn	1 × 262340 × 5376
444	Add_55_attn	Add	1 × 262340 × 5376
445	RMSNorm_55_2	RMSNorm	1 × 262340 × 5376
446	FFN_55	SwiGLU	1 × 262340 × 5376
447	Add_55_ffn	Add	1 × 262340 × 5376
448	RMSNorm_56_1	RMSNorm	1 × 262340 × 5376
449	Attention_56	Grouped Query Attn	1 × 262340 × 5376
450	Add_56_attn	Add	1 × 262340 × 5376
451	RMSNorm_56_2	RMSNorm	1 × 262340 × 5376
452	FFN_56	SwiGLU	1 × 262340 × 5376
453	Add_56_ffn	Add	1 × 262340 × 5376
454	RMSNorm_57_1	RMSNorm	1 × 262340 × 5376
455	Attention_57	Grouped Query Attn	1 × 262340 × 5376
456	Add_57_attn	Add	1 × 262340 × 5376
457	RMSNorm_57_2	RMSNorm	1 × 262340 × 5376
458	FFN_57	SwiGLU	1 × 262340 × 5376
459	Add_57_ffn	Add	1 × 262340 × 5376
460	RMSNorm_58_1	RMSNorm	1 × 262340 × 5376
461	Attention_58	Grouped Query Attn	1 × 262340 × 5376
462	Add_58_attn	Add	1 × 262340 × 5376
463	RMSNorm_58_2	RMSNorm	1 × 262340 × 5376
464	FFN_58	SwiGLU	1 × 262340 × 5376
465	Add_58_ffn	Add	1 × 262340 × 5376
466	RMSNorm_59_1	RMSNorm	1 × 262340 × 5376
467	Attention_59	Grouped Query Attn	1 × 262340 × 5376
468	Add_59_attn	Add	1 × 262340 × 5376
469	RMSNorm_59_2	RMSNorm	1 × 262340 × 5376
470	FFN_59	SwiGLU	1 × 262340 × 5376
471	Add_59_ffn	Add	1 × 262340 × 5376
472	RMSNorm_60_1	RMSNorm	1 × 262340 × 5376
473	Attention_60	Grouped Query Attn	1 × 262340 × 5376
474	Add_60_attn	Add	1 × 262340 × 5376
475	RMSNorm_60_2	RMSNorm	1 × 262340 × 5376
476	FFN_60	SwiGLU	1 × 262340 × 5376
477	Add_60_ffn	Add	1 × 262340 × 5376
478	Output	Output	1 × 262340 × 5376

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 87 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

warnAcross 87 attention layers this design caches 752 KB per token, so a single 8,192-token sequence needs ~6.3 GB of KV cache before weights or activations. That exceeds the 4 GB budget this rule assumes for serving headroom. Fix: Cut KV width: raise the GQA ratio (fewer numKVHeads), switch to MLA, reduce depth or embedDim, or accept a shorter serving context.
kv-cache-context-budget

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 farbodtavakkoli/OTel-2.0-LLM-31B-IT --plan --share

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